Returning to chess after a break: what one year and 188 games taught me

A FIDE Master, I stopped playing chess for six years before coming back in August 2025. One year on: 188 FIDE-rated games, 139 rating points lost at classical, and a far clearer idea of what had survived the break and what hadn't.

The short answer: I didn't really lose my judgement. I lost the ability to calculate properly for several hours at a stretch.

This is my account of that year. Not a theoretical guide: my numbers, my rating curve and what I understood by playing a great deal again. The two summer opens and the team weekend that followed are covered round by round elsewhere on the blog.

Six years without a game — except one

I stopped in 2018. No grand decision, no farewell to chess: I simply stopped playing.

One single game in 2019. Friends needed a player to secure their place in the French second division; I came to help out, played, and left again. One detail matters, because it's the sort of thing you rewrite after the fact: I wasn't planning to come back. Not "maybe one day", not "when I have the time". I wasn't thinking about it at all.

I won it fairly comfortably, with White in a French Defence, against a 2130-rated player. I still remember it well, which is hardly surprising: when you play one game in six years, it stays with you. Here are the twenty-six moves I kept.

The engine, which I ran for the occasion, confirms the memory: no inaccuracies, no mistakes, no blunders, nine centipawns of average loss and 98% accuracy.

I won't make it a proof of anything. Twenty-six moves against an opponent who cracks fairly early isn't a full examination. But it's a photograph of the player I was when I stopped, and it's that player I compare myself to now.

It wasn't my first time drifting away, either. From 2010 to 2017 I had played with no chess ambition at all: to see friends, for the club atmosphere, not to improve. You can sit in front of a board for years without really playing. Stopping in 2018 only made official something that had begun long before.

Then I eased off at work. I had more free time and went looking for a new challenge — not necessarily chess, in fact: I was mostly looking for something to do. Chess was there.

So in August 2025 I went and played the Martigny rapid. Seven games, with no notion of "returning to chess": I wanted to see what would happen. My goal fitted on one line: don't hang a piece in two moves.

I scored 5½/7 and gained three rating points. After six years away, that was encouraging. For a few weeks I thought the comeback would carry on like that.

Do you lose your level during a break?

Not exactly — and that's precisely what the FIDE rating hides best.

A rating doesn't fall while you're not playing. It stays put. My 2289 at classical said nothing about my level in 2025: it described the player I was in 2018.

The problem appears the moment you sit back down. The pairings put you against opponents calibrated for the player you used to be, and the maths still uses your old rating: at 2289 against a 2000-rated player, a draw isn't worth "half a point", it costs you several rating points. I was coming back with today's playing strength and a six-year-old rating. Most of the 139-point gap comes from that.

There's a second, less obvious consequence. Someone who comes back and plays eight games a year keeps their 2289: not because they're still worth it, but because they don't play enough for the rating to show otherwise. I did exactly the opposite.

And on the other side of the board, what's changed?

Chess has changed.

I expected my own game to have shifted after six years away. I didn't expect the game itself to have shifted too.

Six years isn't only six years for me: it's six years for every player I was about to meet again. A 2000-rated player today has little in common with a 2000-rated player in 2018.

First of all, he's a great deal younger than me.

This year I often felt like the veteran at the table. The best example is the last round at La Plagne, in July: across from me, a nine-year-old, right on 2000. Twenty-four years between us. In 2018 he probably didn't know how to play yet.

Six weeks later, in round 6 of the Swiss team championship: fourteen years old, 2313. Nineteen years younger than me and more than 150 points above.

At thirty-three you don't recover from a fourth hour at the board the way you do at fourteen, and you don't turn up to the next day's round in the same state. This isn't a complaint, just a preparation variable I had never needed before.

Second, they have every tool at their fingertips.

In 2018, preparing a line took work. Today the best continuation is a few clicks away, the games database fits in a phone and the engine runs on any computer.

The reversal is amusing: what has become difficult isn't surprising someone, it's playing the main lines. They're far more heavily analysed than they were, and players no longer know only the moves: they know the plans that go with them. An opponent who has forgotten the twentieth move still finds his way, because he knows what he's supposed to do with the position. The variation escapes him; the terrain doesn't.

To get an advantage there, you have to be creative early and turn up with an idea. Two options, and neither is comfortable: play a move the engine rates at 0.00 but which poses a practical problem the other player hasn't prepared, or choose a slightly inferior, lightly analysed line, accepting a small theoretical disadvantage in exchange for leaving the beaten track.

The way a repertoire gets built has changed too.

My serious opening work stopped around 2010. In 2018 I was still playing the same lines and they held up: in several positions they even gave me a theoretical edge over players of my own level. Eight years without touching them says a fair amount about how fast theory was moving back then.

The method was simple. You took a book, read it, entered the variations into a database, looked for ideas and plans. It even happened that you understood the position better than the engine, which already calculated far better than we did but explained things poorly.

Today you open Chessable, find a course, and replay the variations on the board until they stick. And the course is often excellent: up-to-date lines, plans explained, spaced repetition to make it stay. I don't say that with contempt — I use it myself. But the outcome isn't the same: what used to take dozens of hours of personal work, which few players were willing to invest, is now available immediately. Yesterday's competitive edge has become today's baseline.

That obviously doesn't mean opening work is pointless. It depends on your level and the time you have, and since the question comes up often, here's how I see it.

Below 1800, learning the main lines and knowing what to do with the positions that come out of them makes a lot of sense. There's no need to aim for an advantage out of the opening: you need a playable position and the ability to play it.

Between 1800 and 2200, you need to be able to hold the main lines against stronger players, and to keep a few less obvious ideas for weaker ones. Two repertoires for two purposes.

Between 2200 and 2500, the essential thing becomes understanding the structures you play, while reserving a few special ideas for slightly weaker opponents.

At 2150, what I refuse to do is spend my evenings hunting an opening advantage against a player of my own level who has exactly the same engines I do. A great deal of work for very little chance of gaining anything, and I'd rather put those hours into calculation.

Especially since those small theoretical advantages are often hard to play. The engine shows +0.4 and assures you the position is better; you still have to play it. Many of those advantages demand a precision you don't have, in positions you don't particularly enjoy. You can spend fifteen hours earning a +0.4 and then lose because you can't convert it.

Playing style has moved too. Still my fourteen-year-old opponent: he pushes h4, a pawn in front of his own king, without hesitating and at the right moment. That generation grew up with engines and plays without a second thought moves mine would have found dubious, or simply ugly.

And he's much harder to crack.

It isn't that he makes fewer mistakes: the game itself has become more solid. The patterns are known, the general level has risen, the gifts have dried up. In a genuinely complicated position he'll go wrong like anybody else; the problem is getting him there.

The computer hasn't sterilised the game, either — quite the opposite. You might have thought engines would condemn the dubious gambits: they've validated a good number of them. A pawn down, especially with Black, often offers enough compensation to hold the balance provided you play very actively. And they've brought back into play any number of positions we used to consider bad or unplayable.

So the opponent is both harder to surprise and better armed when you do manage to put him under pressure. To beat him you have to create complexity, which means calculating. We always come back to the same problem.

Plenty of them also play for the draw.

Not out of cowardice: it's a perfectly rational strategy. A 2000 who draws against a 2200 has done good business, and he knows it from move one. So he trades, closes the position and waits. He isn't playing to win, he's playing not to lose.

In Spain this summer I had a telling example. My opponent has been a Sicilian player his whole life. We're going over the game together and, somewhere in a variation, he tells me:

"I'd prepared a Petroff to make a draw. There are a hundred points between us, that's enough for me."

So a lifelong attacker had put his Sicilian away in the locker room to bring out the most solid defence in the book. Not for lack of ideas: as tournament strategy. And it was a good one.

That wouldn't come naturally to me. I play to win against anyone, with either colour. It isn't a quality. It used to earn me points; this year it has cost me a great many. But it's how I sit down at a board, and it hasn't changed in six years.

Sébastien Joie, a Swiss international master and one of the nicest people in the game, did make me think about it. I was telling him I'd had enough of players who draw everything. His answer:

"That's my dream, though. I keep the pieces on, I improve my position better than they do, they'll take longer than me to find the right moves and, in the end, I'll make them crack."

He's right. Against an opponent whose first priority is a split point, there's no point hunting for the winning move: you improve your pieces, keep material on, and wait. The clock and fatigue often finish the job.

The trouble is that I could still do that early in my comeback and much less well now. Improving your position for forty moves without forcing anything demands exactly what I'm short of: stamina, and enough confidence in your own judgement to accept doing nothing spectacular.

Then there's the Indian school.

I played ten games against players who come out of it, across all time controls: 3 points out of 10 and 59 rating points lost. In October, in Annemasse, I was still holding my own with 2½/4 and a neutral balance. Since July, in the international opens of the summer: ½/6.

The game-by-game detail says more still. Nine of those ten games were classical, and each one cost me 5.0 rating points on average — against 2.2 versus French players and 1.2 versus Swiss ones, over the same period and the same time control. The sample is far too small to make a rule of it, but the gap is large and it matches what I felt at the board.

I can describe how they play fairly precisely, having been on the receiving end ten times. It's very active, very concrete and very practical: few waiting moves, every move demands an answer. The move chosen isn't necessarily the best one, but the one that will cause you the most trouble given the time on your clock. They calculate fast and well, distinctly better than comparably rated players I've met elsewhere. Their positional understanding seems more fragile to me: long-range plans and the judgement of quiet positions are less impressive. And above all they're extremely combative — a dead position doesn't exist, an equal endgame gets played down to a bare king.

Set against what I've just said about my own game, the problem is glaring: we have opposite flaws. My judgement holds, my calculation is rusty, and I stop fighting the moment I think I'm lost. They calculate better, understand less well, and never let anything go. So I meet them on their own ground — unless I manage to move the ground.

That's exactly what happened in Annemasse. I managed to steer them into positions where there was no tactical decision to be taken: improve a piece, wait, don't force. They didn't stop calculating for all that, and that's precisely where their problem showed — there was nothing to calculate. They went looking for a concrete continuation that didn't exist and ended up forcing matters. Their weakness in those positions isn't independent of their strength: it's the other side of it. That's where my 2½/4 came from.

For the rest of the summer, I couldn't do it any more. Contrary to what everything else in this article might suggest, this wasn't a calculation problem: it was a problem of choosing variations.

The terrain of a game is settled far earlier than you'd think, in the opening and then in the line you go for. Quiet positions don't materialise by magic on move twenty; you often choose them on move five. This summer my choices regularly led me towards concrete positions, and therefore onto their ground.

That's why the repertoire is, of my four building sites, the one moving fastest: I know exactly what to change in it. The remedy isn't to pick quiet positions and stick to them, though, which would mean giving up part of my game. What's needed is to match the choice to the opponent and settle it before sitting down.

Against an older player, I can go for the concrete: it's ground that suits me, and I can lean more on my calculation.

Against a young player, I have to arrive with a plan, and it comes down to two questions: which positions do I think he's comfortable in, and which ones less so. The answer determines my opening and steers my decisions for the whole game, since I know what I'm heading for.

Preparing for a game is therefore no longer just about memorising variations, but about deciding what kind of position it will be played in, depending on the opponent.

It's an uncomfortable situation, coming back with a rating that's still high and a game that no longer quite matches it. Below me, some opponents play for the draw; above me, juniors are very keen to beat the old master back from a six-year break. Either way, you have to create something where nothing is given. That is precisely what I had unlearned.

I can't measure any of this statistically, and I don't want to turn an impression into false precision. But it came back often enough across my 188 games for me to take it seriously. It isn't an excuse: it's information for preparing my games differently. Against someone who has the same tools and makes fewer mistakes, the opening can no longer be where I take my advantage. It has to come later, when the preparation runs out and you have to calculate — which is exactly where I'm least comfortable today.

How much I lost, exactly

It's the last column that matters. The total depends mostly on how many games you play; the average cost per game tells you what happens each time I sit down at a board.

The second-to-last column answers an obvious objection. My rapid and blitz opponents are on average eighty points weaker than my classical ones, and my 75% score at rapid partly comes from that. But the rating calculation already accounts for the opponent's strength: losing 0.6 points a game at rapid against 1.5 at classical is still a real difference.

A hundred and thirty-nine points at classical, put like that, looks like a catastrophe — and that's how I experienced it. Except you have to translate those points into results. My K factor is 20: the points lost are therefore 20 times the gap between my actual score and the score my rating expected.

Across 90 classical games, I scored seven points fewer than expected.

Seven points over an entire year. In other words seven games lost where my rating expected a win, or fourteen wins turned into draws. That's still a lot — at this level, seven points shows — but it isn't the collapse the raw number suggests.

More importantly, I know where those seven points went. They didn't evaporate gradually: they sit in a handful of games I can find and replay.

A piece up on the board, and a mate in two I don't see. A completely winning attack that ends in a draw.

That's what my 139 points look like when you look at the games rather than the rating: a player who often builds a good position and still regularly botches the finish. Nobody took those points from me — I gave them away.

That's rather good news. Points lost from winning positions aren't worked on the same way as points lost because you were outplayed from start to finish. My game hasn't disappeared; the finishing has. For comparison: at rapid, across 67 games, the deficit is only about two points.

Hence the first piece of advice I'd give anyone coming back and watching their rating slide: translate the loss into games. A falling rating makes you feel you've forgotten how to play; the same gap expressed in half-points tells a story much closer to what actually happened at the board.

What comes back, and what doesn't?

Take the last column of the table again: 1.5 points lost per game at classical, 0.6 at rapid. A classical game costs me two and a half times more than a rapid one, and I'm far closer to my old level today with twenty minutes than with four hours.

That's counter-intuitive — more time ought to help me — and it's one of the most important things I understood this year.

What comes back fast is judgement. Recognising a structure, knowing which piece is badly placed, understanding what you're looking for, spotting a weakness: none of that had gone, it was dormant. Rapid is largely played on those qualities, which probably explains why it held up.

What comes back with work is tactical sharpness. Early in the comeback it had gone too: I was seeing some patterns too late, or not at all. The good news is that it's coming back well, and I know what's doing it — volume, tactics almost every day, Puzzle Storm above all. It's repetitive, not very glamorous, but it works.

What comes back much less readily is calculation. And it must not be confused with tactical sharpness. Sharpness is seeing that there's something there. Calculation is following the line for six to eight moves, coming back, checking you haven't missed anything, then trusting it enough to play it. The first is trained by volume; the second takes far longer, and I've found only one exercise that attacks it head-on — I'll come back to that at the end.

I can compare precisely, because I know what I was worth before.

Stamina is the easiest line to explain, but the last one is a better thermometer.

Positions where there's nothing to understand — a small endgame, three pieces each, every piece of information on the board — used to be my territory: I calculated everything, often better than my opponent. Today that's exactly where I go wrong.

Round 7 of the Swiss team championship against Cédric Pahud is a good example. I play 41...h5, a move I call atrocious myself in that diary: it loses a pawn in three moves. My opponent doesn't see it. Three moves, in a position where there was practically nothing to do but calculate. And the final half point I didn't find either: my opponent handed it to me.

Early in my comeback, the pieces would literally move around on their own inside my calculations. I'd start a line, go six or seven moves deep, and on the way back the mental board was no longer the same: a bishop had changed square, a pawn was still sitting where it no longer stood. The image came apart, and I spent checking I hadn't misread it the time I should have spent calculating. When you no longer trust your own mental board, you start over — and in the fourth hour, recalculating everything is very expensive.

You could see it on the clock. I was permanently in time trouble back then, not because I thought too long about each decision, but because I kept recalculating what I'd just calculated. I was playing with very little confidence in my own variations.

I managed to fix that flaw, and the time trouble has all but disappeared: I reach the move-forty control with something in hand, which wasn't the case in the first season. I've simply swung to the other extreme — I no longer calculate far enough.

My two current flaws are quite specific, which is good news in itself: a specific flaw is easier to work on than a vague sense of no longer knowing how to play.

I sometimes rush my variations. I stop a move too early, without going through the check, the capture, the threat. That's exactly the small verification you should run at the end of every line; it takes no intelligence, only discipline.

And I get lost in complicated positions. Kotov, in Think Like a Grandmaster, describes calculation as a tree: candidate moves, branches, and the constant risk of going over the same ground twice. When the line is long but forced, with a single branch, I still handle it well. When the position offers four or five candidate moves at every level, I lose my way: I set off down a branch I've already explored, forget another one, and end up deciding on feel.

So it isn't the calculating power that has gone, but its organisation — the automatism, more simply. I still know how to calculate; I no longer calculate naturally.

My tastes, on the other hand, haven't shifted a millimetre. I used to love concrete positions and I was good at them. Today I love them just as much and I'm less good at them. That explains part of my summer troubles: the ground my opponents are most comfortable on is exactly the one I head for by instinct, because it's the one I enjoy.

I don't deduce this from my statistics alone; I saw it during the games and found it again in my own diaries. At the Swiss team championship I had found the winning plan — trade queens, bishop to b4, rook to c2, then push the majority. I didn't play it. The next day I had another good idea and didn't play that one either. Twice the judgement was there; twice I didn't calculate enough to trust it.

And rapid, then?

Rapid is the only one of the three formats where I still recognise my old game. So it deserves a closer look.

67 games, 50 points: a 75% score, for −40.6 rating points. The paradox isn't one: at 2285, winning three games out of four against opponents who are on average weaker isn't enough to hold your place, you'd have to win more. But the average loss is only 0.6 points a game, against 1.5 at classical. That's a great deal less.

It's also the only format where I've had genuinely good days: +11.4 at Le Bouveret in February, +7.2 at the Jura Open. I've had bad ones too — −19 at the Perolini Trophy — but they didn't turn into a trend.

I see two reasons for it.

The first: I have good intuition and poor resistance to fatigue. In a twenty-minute game only the first is called upon; the second doesn't have time to come into play. The fatigue doesn't disappear for all that — a rapid tournament runs a whole day and I often fall apart towards the end of it — but it arrives later, and the speed brings an adrenaline that outlasts my concentration.

The second: at rapid I'm tactically sharper than a lot of my opponents. I see the patterns fast, which shows the problem isn't recognition — that's exactly what Puzzle Storm repaired first. What stays hard is long calculation: holding a line over six to eight moves, coming back, checking, not losing the thread.

Rapid therefore calls on the part of my game that still works. It shields me from my weakness and shows off what I have left. That's pleasant, and it's a trap: the rating I care about most is still the classical one.

And blitz, then?

Blitz shows −68 points, the worst per-game deficit of the three time controls. I don't draw much from it, and I'd rather say so.

To begin with there are very few rated blitz tournaments: my 31 games for the year fit into three afternoons, one of which accounts for almost the whole deficit. Blitz is also often played alongside another tournament — the La Plagne one was held on 9 July, in the middle of a nine-round open running from the 4th to the 10th. You turn up already worn.

Above all, blitz ratings are often old, and the mechanism looks a lot like mine, in reverse. Since there are few rated blitz events, a player only gets two or three a year: his blitz rating barely moves. Meanwhile he improves — he plays online, works on his tactics, gets better. The number stands still, the player moves on. Two years later you sit down across from a "2000" who is worth considerably more.

My rating stayed high because I wasn't playing; theirs stays low because they have no chance to move it. In both cases the number no longer describes the player — except that mine was too high and theirs is too low. A blitz rating therefore also measures how often tournaments are held.

One more story pointing the same way: it was in a blitz tournament, precisely, that I beat a grandmaster this year. The same rating that shows me at −68 also contains that game. Better not to make it say more than it can.

How long before the rating settles?

Here's the curve, month by month, at classical.

And the same figures, for anyone who prefers to read them:

Three things surprised me when I looked at that curve.

In the first month, I went up. In October 2025 I climb to 2292. The rating didn't give way on contact, it even rose slightly — and of course I believed it.

Then almost all of it fell in four months. From November to February, 2292 → 2171: 121 of the 139 points lost over the year go in that one window. All the rest of the year is worth eighteen points.

Then five months of stability. From March to July I stay between 2165 and 2170. Since I was playing regularly, I ended up believing I'd found my level. I was wrong.

The summer reminded me the curve wasn't finished: La Plagne cost me 17.8 points, Badalona 14.6. The team weekend that followed isn't even counted yet, and September is already announced at −15.2, which would put me around 2135. The plateau wasn't my final level, then, merely a period when the rating had stopped moving.

The honest answer to this section's question is therefore: I don't know. After a year and 188 rated games, I still don't know where my rating will settle. I thought I knew in June; I was proved wrong in August.

If you're coming back after a long break and your rating holds steady for a few months, I'd be careful about concluding anything from it. A plateau looks a lot like a floor when you're standing on it, and you only tell the difference afterwards.

Not knowing where it stops doesn't mean there's nothing to be done, though. A rating that falls for reasons you can't name is worrying; mine falls for a reason I can name. I don't calculate well enough yet and I don't always trust the lines I find. That's a specific weakness, and a specific weakness can be worked on. So the first thing that has to come back up isn't the rating: it's the play. The rating will follow.

The year, tournament by tournament

Here's the full year. I'm giving all of it, because an account that shows only the good tournaments is worthless.

And the team competitions, grouped together because they run across the whole season:

This table shows things the curve doesn't.

Blitz is played in half a day. On 4 October 2025, at the Swiss blitz championship, I score 8½/13 — a perfectly decent result — and lose 59 rating points. Almost my entire blitz deficit for the year comes from that single day, which is enough to stop me drawing any general conclusion from it.

The big losses come mostly from opens where the opponents are solid. Annemasse in October and Vandœuvre between Christmas and New Year cost me 55 points between them. Going looking for strong opponents when your calculation hasn't come back has a price.

The worst tournament of the year is still the group championship, though: half a point from five games, −26.6. Five team games spread over five months, sometimes a fortnight apart. It's my most concrete lesson of the season — I play badly when I only get one game every two weeks.

There are two pieces of good news I did tend to forget: Le Bouveret in February, +11.4, and the Geneva Open in May, +8 with 5/7. Two tournaments where I genuinely gained points.

One last figure could pass for a bright spot without being one. Pontevedra only cost me 0.6 points; my first instinct was to read it as a good sign. But a balance close to zero simply means I played at the expected level that day, nothing more. The proof: the following week I lose 14.6 points in Badalona, then play a team weekend straight after. Part of my problem is probably right there — I can't yet hold my level from one tournament to the next.

What I'd say to someone hesitating to come back

Play rated games again quickly, and play a lot of them. A rating that hasn't moved in six years isn't a level you've earned: it's a number that hasn't been tested. The longer you wait, the wider the gap between the number and the player. My 188 games at least let me find out where I actually stand, and you can't build anything on a rating you don't believe in yourself.

Start with rapid. I did it without deciding to and it was the right order. Rapid lets you lean on what still works, whereas a classical game exposes what's missing straight away. Coming back with four-hour games means going looking for your weak spot on day one.

Separate judgement and calculation in your training. They're two different qualities, and they don't come back at the same rate. Working on "chess" as a block doesn't necessarily repair the one that's failing. In my case judgement had almost nothing to repair; calculation had to be rebuilt from scratch.

Don't check your rating every Monday. Over twelve months mine went up, collapsed, flatlined for five months, then fell again. Only one of those movements taught me anything about my play; the other three mostly cost me morale.

I have no comforting moral to offer. I've lost 139 points and I probably haven't finished losing them. But I now know what I have to work on, what my 2150 is really worth and why it doesn't look like my 2289 — and I would never have known that without coming back.

I'm not promising I'll see 2289 again, I have no idea. What I do know is that a decline whose cause you understand isn't a fate: it's a problem to work on. In my case there are four of them.

The four building sites

The repertoire. The furthest along, almost rebuilt. My old lines kept evolving while I wasn't playing, some have been refuted by the engine, and the ones that still hold are far better known to everyone else. So a large part of it had to be redone. It's also the only site I can work on almost entirely away from the board, which probably explains why it's moving fastest.

Calculation. Calculate a lot, work on tactics a lot, and above all recover the habit of following a line to the end. The aim isn't to become cleverer, but to get the automatism back. I haven't found a shortcut.

Stamina. The numbers are clear: 1.5 rating points lost per game at classical against 0.6 at rapid. Between the two there isn't three and a half hours' difference in my understanding of the game — there's three and a half hours more concentration. I used to string four or five hours together in a day without thinking about it.

Endgames, rook endgames above all. That's where I've left the most identifiable points, and it's also where the context has changed most. With the increment, a rook endgame is no longer adjourned and no longer ends on the flag: it can run for a very long time at thirty seconds a move. At that rate it's no longer enough to be able to calculate it, you have to know it — whoever recognises the right plan immediately gains an enormous amount on whoever has to work it all out. For now I'm in the second category.

These four problems all meet at the same place: the fourth hour, when there's still an endgame to calculate with whatever concentration is left. That's probably where I've lost the most points; so that's where I'll be going to get them back.

How I train

Concretely, here's what I've been doing for a year. Nothing sophisticated, and that's deliberate.

Volume on tactics, almost every day. Puzzle Storm for speed and pattern recognition: it's exactly suited to my problem, since it requires recognising the solution instead of recalculating everything. I also use the Woodpecker method, which means going back through the same sets of tactics until the solutions become familiar. The idea appeals to me because it targets automatism rather than understanding — and automatism is precisely what I'm missing.

Studies, with John Nunn's Endgame Challenge. It's probably the only exercise that genuinely improved my calculation this year, and I think I know why. A study is a stripped-down position where there's nothing to understand: no strategic plan to bail you out, a single solution, often a long one, which has to be found before you play the first move. That's exactly what I can no longer always do in a game.

Chessable for the repertoire, mainly for the spaced repetition rather than the courses themselves. The review system does what an adult player doesn't do well alone: come back to a line at the moment he's starting to forget it. Rebuilding a repertoire at thirty-three isn't a problem of understanding but of memory, and on that point the tool is more reliable than my willpower.

None of this is original, and I'd be wary of a method that claimed to be. They're simply exercises matched to the problems I've identified.

The interesting part starts now.

If you're in the same situation and want to structure your return to competition, that's exactly what I work on with the adults I coach: see adult chess lessons in Geneva.