Bet Size as the Line Between Courage and Recklessness: the Kelly Criterion, Ergodicity, and Two Emigrations Without a Plan B Author: Дністер Published: 2026-09-13T03:01:33.000Z Language: en URL: https://neurodrift.org/en/blog/vystryb-u-nich/ Original (Ukrainian): https://neurodrift.org/blog/vystryb-u-nich/ Tags: risk, identity, decisions, psychology Original source: https://neurodrift.org/blog/vystryb-u-nich/ Twice I leapt into the dark with no floor under my feet: Britain in 2012, Africa in 2019. Between them, the depression of 2014 — which taught the one difference between a daredevil and a disciplined operator: the size of the bet. The Kelly criterion, Peters' ergodicity, survivorship bias, gambler's ruin — and my own documents from Drive. ----- "The deed of the leap into the night — deciding to go despite a tiny sum of money, betting that Abdul will add money and help with lodging." — my own document, 2018 I. 60% — and why you can't bet it all In 1956 a Bell Labs physicist named John Kelly derived a formula that answers one simple question: if you know for certain the odds are in your favor, how much should you bet? Intuition screams: everything. If I'm at 60% to 40%, pile in — it's an obvious robbery in my favor. Kelly's math answers coldly and insultingly: 20%. Not a hundred. Twenty. Because whoever bets it all on a positive edge doesn't get rich — he goes broke. Just with better statistics than a fool. I didn't learn this formula from a textbook. I learned it by smashing into the asphalt in 2014. And then, five years later, in a dusty jeep between Nairobi and the Tanzanian border, I made the same leap correctly for the first time — and didn't immediately realize it was Kelly's formula, rewritten in sweat and dust. A leap into the night looks like courage. It's actually a bet. And a bet has a size — and this whole text is about how I learned to calculate it: first by breaking, then by surviving. II. The planes that came back Before we get to me — about why you believe in leaps at all. Because you believe in them on bad data. World War II. American bombers return from raids full of holes. The military wants to add armor and studies where the holes cluster most: wings, fuselage, tail. The logic is obvious — armor where it gets hit most. And then the mathematician Abraham Wald, of the Statistical Research Group, asks the question that chills you: where are the holes on the planes that didn't come back? Because the sample is made only of those that flew home. The engines on the returning machines are nearly clean — not because engines weren't hit, but because the ones hit in the engine are at the bottom of the English Channel. The armor belongs precisely where the survivors have no holes. The data is filtered by survival. What you see isn't the truth about the battle. It's the truth about those who survived for you to see it. Every inspiring story about a bold leap is a bomber that came back. You only see the planes that flew home. The ones that fell with the same drive and the same certainty don't write threads. The whole hustle culture — "just leap," "fortune favors the bold," "burn the boats" — is assembled from interviews with those lucky enough to make it home. Founders who bet everything and won give podcasts. Founders who bet everything and lost stay silent — they're not in frame. You calibrate your own courage on a sample with all the dead carefully cut out. III. One life, not an ensemble Survivorship bias tells you what you see wrongly. But there's a deeper layer that tells you why it's lethal. It's called ergodicity, and it's the most important idea you were never taught. In 2019 the physicist Ole Peters published a paper in Nature Physics that quietly undermines half the popular advice about risk. Imagine a game: we flip a coin. Heads — your capital grows 50%. Tails — it falls 40%. Even odds. Will you play? The classic answer: of course. The expected value is positive — on average you win ~5% per round. But that's the average across an ensemble: if a thousand parallel versions of you played at once, their combined capital would grow. The trouble is, a thousand parallel yous don't exist. There's one of you, playing round after round in time. And the time average for a single trajectory is minus ~5% per round. Almost every individual player slowly slides toward zero while the "ensemble average" proudly climbs. Hustle culture sells you the ensemble average — the success of a thousand parallel versions of you. But you live one trajectory, and in it ruin is an absorbing wall: fall to zero and no "average" will resurrect you. Here's where ergodicity stitches into Wald's planes. The survivors you see are the lucky paths of the ensemble. The podcast founder who bet it all and won is one trajectory in a thousand, and it's exactly the one they show you. You don't play for the ensemble; you play for the single trajectory you have — and in it ruin is absorbing: so leap big, but never with everything, because from zero no average will ever reach you again. !One lit figure walks a solid path forward while around them faint translucent versions of the same person take divergent routes that fade away. You see the ensemble average — a thousand parallel versions. You live only one trajectory. And in it, zero is absorbing. And here you can see the difference between the two people I was. The daredevil bets as if he has a thousand lives — he plays for the ensemble. The disciplined operator knows there's one life, and he plays for the time average. Between those two — exactly one broken year. IV. "The deed of the leap into the night" "The deed of the leap into the night" isn't a metaphor for beauty. It's an internal term I once wrote down for myself. The structure is always the same: you leap before you have all the resources. You don't wait for the floor to assemble under your feet. You leap into the dark betting the floor will appear mid-flight, or that you'll make it from what you brought. The Norse berserkers threw themselves into battle without armor — naked, in a trance. Half won so completely that sagas were written about them. The other half simply died, and nothing was written, because there was no one to write it. The difference isn't fury. The fury is the same. The difference is how much each one staked. And that you hear the saga, but not the silence. V. Britain, 2012: a plane that came back I was twenty. I packed and left for Great Britain — no visa, no plan B, no return ticket in my head. The pretext: there was a girl I was going for, and her builder father had work for me. The substance: I leapt into a country where I was nobody, an illegal with no status, betting the situation would somehow hold. And it held. Twenty-two months. Physical labor, sixty-hour weeks, a body that after a shift wanted only to lie down. Not romantic emigration from Instagram — wet concrete, early dark, a foreign language. But over those months I saw fourteen British cities: from Edinburgh and Glasgow to Brighton, Swansea and Loch Ness. I'd never have gone anywhere on my own — she shoved me out, with kicks. And it was that absorbed confidence — that you can just up and be somewhere else — that became the real cargo I brought back. Not money. Optics. The leap to Britain didn't give me money. It gave me proof that I could land in a foreign dark and not vanish. That proof later cost more than any wage. That's what a leap into the night looks like when it works. And here's the trap: twenty-year-old me drew the wrong conclusion. I decided the leap itself was the victory. That the main thing was fury and readiness, and the floor would always appear, because last time it did. I became a bomber that came home and decided it was invulnerable. That's the gambler's error — raising the stake after a win. !A young man in work clothes against a wet British construction site at early dusk. Twenty-two months of wet concrete. Britain paid not in money but in optics. And it nearly sent the bill the next year. VI. 2014: the night with no floor I came back to Ukraine and fell out. Not metaphorically — clinically. What the archive would later honestly call a depressive episode. The despair was fueled by the absence of effect, of significance, of being needed. The leap was over, the adrenaline burned off, and the floor you can stand on day to day — I hadn't brought it. I knew how to leap. I didn't know how to land. On 6 October 2014, at 22:41, I created a file in Drive: "100 Days to Death." I declared: I'll write a novel, a hundred days straight, a chapter a day. The epigraph ended in bold: "Otherwise I die unwritten." Of the hundred declared days I wrote five percent. A full Day One, headings for days 2–5 with no text — and that's all. The hero of that abandoned novel is Plumbum, with a prosthetic arm rusting in the running water of his own apartment. He isn't a literary hero in an extreme state. He was me, covered by a thin layer of the absurd — thin enough not to recognize myself. And the Kipling pastiche I ended not with the promise of becoming a Man, but with the promise of becoming a Madman — amid "stupidity, disability, and numerous suicides." I declared a hundred days. I wrote five. That's a result too — it's just called not "a novel" but "a diagnosis." And it's the most honest document I have about myself. In 2014 I bet as if I had a thousand lives — I played for the ensemble. The first leap had external structure (work, a foreign country, a girl-as-engine); the second was a pure leap into the void with no valve. The same fury. The same berserker. Just without armor — and my single trajectory came right up against the absorbing wall. I didn't hit the bottom only because I was pulled out — not because I'd calculated the height right. And that's exactly what I'm obliged to remember every time the next leap starts singing in my blood again. VII. The valve Four years passed. On 21 December 2018, a week after being fired from the Kyiv Post, wintering in Thailand, I wrote a document: "My Goals to 35." Among them — the formula "Prove to myself I'm a hard, aggressive Berserker," and its last line: "…go for powerful risk and don't be afraid to start from scratch (stay hungry). But diversify, so the depression doesn't hit hard." Two sentences that argue on purpose. The first — berserker. The second, in fine print: diversify, so the depression doesn't hit hard. That's a man who has already crashed and remembers exactly how. Twenty-five-year-old me didn't know he was quoting both Kelly and Peters at once. But he was doing exactly what they derived: stake a large fraction, never the whole bankroll, because you play your single trajectory, not the ensemble. "Diversify, so the depression doesn't hit hard" is Kelly's fractional bet, only the unit of measure here isn't dollars but the psyche. PlayerWhat he plays forFate at a positive edge CowardDoesn't bet — waits for the floorDoesn't go broke, but doesn't grow. Dies of the plateau Daredevil (me in 2014)For the ensemble (as if lives are many)Goes broke almost surely — even with an edge Operator (me since 2018)For the single trajectory in timeGrows over the long run. Falls hurt but don't kill The market confirms this with a number that runs cold in your gut. J.P. Morgan analyzed every company in the Russell 3000 from 1980 to 2014: about 40% of stocks suffered a "catastrophic loss" — a 70%+ decline with no recovery; two-thirds lost to the index itself. Even if you pick winners in a market that rises on average, putting everything on one position is a 40% chance of zeroing out forever. The market itself screams: diversify or die. The night has no floor. You bring the floor. In 2012 I brought it by accident. In 2014 I left it at home. Since 2018 I pack it on purpose — and call it "diversify." VIII. Africa, 2019: the same leap, but with a parachute Seven years after Britain I did it again. Two months wandering with my brother through the backcountry of Kenya and Tanzania. The formal pretext — combine recreation with work meetings with a client, Abdul. The real structure — the same leap: to go despite a tiny sum of money, betting that Abdul would add to it and help with lodging. The twenty-five-year-old daredevil would have done it blind, dumping the whole bankroll, and prayed. I in 2019 did it with discipline — the difference isn't the sum but the architecture of the bet. Because this time several nets were strung under the leap: Not one bet, but two on one ticket. Recreation plus work meetings. If the business part fails, the trip remains, and vice versa. Diversification right inside the trip. Behind me — a company, not a void. In 2012 failure meant zero. In 2019 — a fall onto a working infrastructure that generates money without me. A leap from a foundation is a different leap than one from the edge. I didn't stake the whole psyche. The main lesson of 2014: the most fragile asset isn't the wallet but the nervous system. This time I knew where the valve was, and kept my hand on it. The leap looked just as romantic as at twenty. An Instagram tourist would see the same adventure. But the difference between 2012 and 2019 is the difference between a berserker who runs into battle naked and one who has invisible chainmail under his skin. The fury is the same. The survival is not. !A dusty jeep on a dirt road in the East African savanna at sunset, two silhouettes inside. Kenya–Tanzania, 2019. The same dark, the same "tiny sum." The only difference was the chainmail under the shirt. IX. Before you step: calculate the height You almost certainly have your own leap into the night: the moment you threw over the stable for the uncertain. And you remember it one of two ways — either as the best thing that happened to you, or as the thing after which you spent a long time gathering yourself in pieces. The difference is almost never in the size of the dream. It's in the size of the bet. Here's the checklist that cost me one broken year: QuestionBad leap (for the ensemble)Disciplined (for your trajectory) What's under the leap?A void. Failure = zeroA foundation where falling isn't fatal Am I staking the psyche?All of it, all-inNervous system in reserve, untouched How many points of support?One: "it has to work"Two or three independent (recreation + client) Where's my Abdul?All hope on one personThere's a plan if Abdul doesn't come through Whose example inspired me?A plane that came backI counted the ones that didn't, too Leap. Seriously — leap. The plateau kills more slowly than failure, but more surely: the coward dies of unrealized life as reliably as the daredevil dies of bankruptcy. The operator-by-Kelly is the only one of the three figures that grows over the long run. Just before you step into the dark, calculate the height and pack the parachute — so that after the landing, someone is left to tell the story. X. In place of a conclusion The romance is in the leap. The discipline is in the size of the bet. Everyone sees the first. No one sees the second — and it's exactly what decides whether there'll be a saga about you, or silence. And the last, most unpleasant truth. You've read one more story about a leap that worked — mine. I'm a bomber that came back too. The only reason this story doesn't look like the rest of the hustle-porn is that I showed you the hole in the engine too: 2014, a hundred days, five percent, the Madman. Most narrators cut that part. So remember not my leap but my valve — and what separates a daredevil from an operator: the daredevil plays as if lives are many. The operator knows there's one. A berserker who didn't calculate the height isn't a hero. He's a corpse with a good story no one's left to tell. A berserker who packed a parachute looks just as reckless at the moment of the leap — and meets the morning. The whole difference between a saga and silence is in how much you staked. And you don't hear the silence only because the dead don't run podcasts — and there's no ensemble left to average your single life back into the black. Sources & context Kelly, J. L. (1956). "A New Interpretation of Information Rate," Bell System Technical Journal. The 60/40 → bet 20% example — Survivorship bias: Abraham Wald and the Statistical Research Group, WWII — put the armor where the surviving planes have no holes — Peters, O. (2019). "The ergodicity problem in economics," Nature Physics. Multiplicative coin toss (+50%/−40%): the ensemble average grows ~5%/round, the time average of a single trajectory falls ~5%/round — Gambler's ruin: raising the stake after wins without cutting after losses → inevitable ruin even with a positive expectation — Cembalest, M. / J.P. Morgan (2014). "The Agony and the Ecstasy." Russell 3000, 1980–2014: ~40% of stocks a catastrophic loss (−70%+ no recovery); ~2/3 lost to the index. Personal archive: "My Goals to 35" (Google Doc, 2018–2025) — the Berserker formula and the line about diversification; the block on Africa with my brother and Abdul. "100 Days to Death" (Google Doc, 2014-10-06) — a declared novel, ~5% written, a Kipling pastiche ending in "you'll become a Madman." The Havryliv interview (2014) — the British period, 22 months, 14 cities.