Isochrons

An isochron is a line joining points of equal time.

The same root gives isochronism: a pendulum takes the same time to swing whether the swing is wide or narrow. Galileo is said to have noticed it watching a lamp move in a cathedral. Clockmakers spent two centuries chasing it, because a clock is only as good as its ability to ignore how far the pendulum travels.

We named the firm after that, not after anything the market does. Same size, same target, same forced exit, whether the week moved 4% or 40%.

Our signal is not a secret. Short-term mean reversion is the most written-about trade there is, printed in open collections, and we would rather say so than dress it up.

What is rare is doing it the same way in a calm week and a wild one. Every time we broke our own rules by hand, it was in a week that felt like an exception at the time. That is the whole claim, and the rest of this page is the evidence for it.

Our strategy has three settings: where it takes profit, how long it will sit in a losing position, and how often it may enter. Every combination those settings allow is one configuration, and there are 315 of them. We tested all 315 rather than the one we liked.

Each stroke below is one configuration. The tall one is the setting we actually trade. It sits in the middle of the set, not at the front: had we picked the best-looking one instead, this page would read better and the account would do worse.

We short ETH against our own ETH. The stack sits in ETH and the short is opened through a contract that settles in ETH rather than in dollars, so what we put up, what we make and what we lose are all ETH. The account grows in coin, and nothing has to be converted back.

One thing follows from this, and it matters: we can never end up short more ETH than we own. The stack behind the trade is the ceiling.

Every ETH in this account belongs to the person who runs it. There are no outside investors. Isochrons Capital is the name the work is published under.

We would rather say that plainly than leave it to be assumed. It is also the strongest form of the claim: the rules on this page are followed with the money of the person who wrote them, and there is no other money to follow them with.

Sergey Kalinin.

The paragraph above is worth what the person in it is worth, so here he is on the same terms as the rest of the site: dates and facts, no adjectives.

2008. I started trading through the mortgage crisis, on the short side, alongside a day job analysing commercial property investment. The short side was a beginner's choice and it has not changed since.

2016. Into crypto.

2017. I shorted BTC against dollars rather than against a stablecoin. The distinction looked pedantic at the time. It is the reason this account denominates a short in the asset it is short of, which is the first rule the strategy is built on.

2021. A short of mine was caught in a sustained rally and liquidated. The position was right about direction and wrong about duration, which is the specific way this class of strategy fails.

2024 to 2026. The reason the short side interested me, that its volatility is structurally higher because spot can only ever be long, stopped being a belief and became something measured. Nine years of data, a forced exit, and a test that charges for every configuration tried.

The rule that cuts a stuck position after about four weeks was not chosen because it flattered a backtest. It is 2021 written as code, and it carries the largest measured term in the result.

Twenty years of deciding what an asset is worth and what it will do next. Performance Officer at a consultancy in 2005, where the job was to build the first set of performance metrics the firm had. Investment analysis at Colliers International through 2008 and 2009. Country Managing Director of Conwert AG from 2009 to 2013, a listed investment holding with operations across Austria, Germany, Hungary and Ukraine, running the country operation and answering for it. Private equity and sell side advisory from 2013 to 2017. Crypto from 2018 as a product lead: an OTC desk, then Head of Product at the KUNA exchange, then four years of blockchain engineering at DataArt.

The first job on that list was measuring performance, and this is the same job. What changed is the asset. A building is valued by working out its best use and what a buyer will pay for it, a company by what its cash flows survive, a strategy by whether its record holds up once you charge it for every configuration you tried. The tools differ. The discipline is one discipline, and the failure is always the same failure: believing a number because it flatters you.

Which is the other half of the name at the top of this page. A pendulum keeps its period whether the swing is wide or narrow, and that is the property a clock is built around. Twenty years, four asset classes, one question asked the same way each time.

There is no fund here. Nothing on this site is an offer, there is no mandate to hand over, and no client money has ever been traded on this account. That is stated again on the access page, and it is the same sentence in both places on purpose.

It is not modesty about size. Something specific is in the way and it is unsolved. The short is covered by the coin sitting behind it, and that cover is what makes a rising market survivable rather than fatal. Anyone able to withdraw that coin turns a covered short into a naked one retroactively, without a line of code going wrong, and the code cannot yet shrink a position when its collateral walks out. We measured that failure before looking for a way around it. Until the automatic reduction is written and tested, holding somebody else's coin would mean running a risk on their behalf that we have declined to run on our own.

If this ever becomes something other than one account trading its own money, it will be co-ownership rather than a product, and it will start from a conversation with a particular person rather than from a web page. That is a description of how we work. It is not an invitation, and there is nothing here to subscribe to.

Three things were taken out of the strategy after they failed, and each cost us months to admit.

Averaging down. Over nine years at identical exposure, single entry returns 11.45× against 1.96× for the pyramid. It sounds reasonable and it feels brave. It is neither.

The entry pattern. We read candles and looked for signs a move was running out of steam. For a year. When we measured it, it no longer mattered. The market does not care how good our reasons sound.

The second asset. We ran a symmetric leg on bitcoin and put it through the same audit ETH passed. Across 405 versions the result could not be told apart from luck: 83% of the time chance alone would have done as well, against 2.8% for ETH. What that test asks, and why a good backtest is not evidence on its own, is written up in the notebook. The leg closed on 1 August. One account, one asset, from here.

We will fall behind in rising markets. We take only about 0.8 of the market's move, and everything we earn on top of it comes from the falls: we beat the market on 27% of up days and on 72% of down days. In a long bull run we will look silly. That is the price of the structure, not a fault.

Drawdowns will be deeper than anything in the reporting so far. Measured over nine years at the size we run now: 34.8% in coin terms.

There is no stop-loss in the usual sense. A position closes at its target or on the clock. That is on purpose, and it is exactly why we keep spare capital and stay well under the size we could take.

Every idea we tested and dropped goes in the notebook next to the ones that passed. Deciding not to launch takes the same weeks of work, and the list of what we threw out says more about us than the list of what we kept.