A Position-Sizing Framework That Survives Bad Years
TL;DR
- Expected return tells you whether a strategy is worth running; worst-case drawdown tells you how much of it you can afford to run — size off the second number, not the first.
- Staying in the game has two floors: your own tolerable dollar loss, and the capital a strategy’s economics need to keep working through a bad stretch.
- Published drawdowns and fee-coverage figures come from backtests, so bad years can exceed them — treat both as floors and build in margin.
Size for the Hole, Not the Headline
Almost every sizing conversation I have with newer systematic investors starts from the wrong number. They ask what the strategy returns on average and scale capital to capture that. It feels rigorous, but it sizes you for a year you will rarely live in. You never experience the average year. You experience the sequence — up 20%, flat, down 25%, up 12% — and by the time the bad year actually arrives, you have long forgotten the average you sized for.
The failure mode is predictable. A position sized for the good years loses more than planned in the bad one. The loss breaches whatever pain threshold you quietly held, so you cut the position near the bottom, skip the next rebalance, or abandon the approach entirely. The strategy never gets a fair trial because you were never actually invested through a full cycle. Sizing mistakes rarely look like sizing mistakes when they happen; they surface as “this strategy doesn’t work,” when the strategy was fine and the size was wrong.
The fix is to reverse the order of the questions. Expected return — realistically, the edge net of costs that a backtest can defend — decides whether a system is worth running at all. Worst-case drawdown decides how much of it you can afford to run. Those are different numbers, and the second one is the binding constraint. A framework that survives bad years sizes off the worst case first and treats the return forecast as a secondary input: an input to strategy selection, never to position size.
The arithmetic underneath is simple. Your allocation ceiling equals your tolerable dollar loss divided by the assumed worst drawdown. If losing $5,000 would hurt enough to break your discipline, and a strategy’s worst documented case is a 30% drawdown, the ceiling on your position is about $16,700 — no matter how appealing the average year looks. Raise the assumed hole to 45% because you doubt the backtest, and the same tolerance caps you near $11,000. That single formula forces you to answer the honest question first: how much can I lose and still keep following the rules?
The Math of Staying in the Game
Underneath that formula sits the asymmetry that makes drawdowns expensive. Losses and gains do not cancel at equal size. A 30% drawdown needs a 42.9% gain to get back to even; a 40% hole needs 66.7%; a 50% hole needs 100%. The deeper the hole, the harder the climb, because every subsequent gain compounds on a smaller base. This is not theory you can outwait: an account that falls 50% must double just to stand still, and meanwhile the calendar keeps running.
Staying in the game therefore means clearing two separate floors, and your size has to respect both.
The first floor is your own tolerance. Decide the deepest hole you could watch without your process breaking — without cutting at the bottom, without skipping rebalances, without lying to your records. If you cannot honestly survive a 30% hole, then no expected return justifies a position that can fall 30%, because a position you will abandon at the low has no expected value at all. Your tolerable loss is a behavioral fact about you before it is a number in a spreadsheet.
The second floor is economic, and it is the one most self-directed investors miss. Every strategy has a capital level below which its economics stop working — where fixed costs consume the edge and the system quietly becomes a drag instead of an engine. A strategy that only functions above a certain account size will betray you precisely in the bad years, because that is when your capital shrinks toward the floor. Sizing is not complete until you have checked that your allocation, after the worst case you assume, still lands above the strategy’s economic floor. Then add margin, because backtests are optimistic and bad years routinely exceed published drawdowns.
Matching Capital to a Strategy’s Economics
This is where fee-coverage figures become a genuinely useful sizing input, and Kairos Trading is the cleanest published example I know. The publisher runs four systems on a flat $100/month subscription each, and for every system it publishes a minimum-capital estimate: the portfolio size at which the strategy’s backtested excess return versus its benchmark roughly covers the fee. Those figures are fee-coverage estimates, not required minimums — nobody is telling you that you cannot run the system below them. They are telling you where the fee stops being noise and starts eating the edge.
Read them as floors, and the lineup is remarkably instructive. Leader Rotation, the flagship monthly momentum rotation, lists $16,000 with a backtested maximum drawdown of 6.7% — shallow enough that a modest allocation stays above its fee-coverage line even in its worst documented stretch. QQQ Top Stock Rotation lists $21,000 against a much deeper 29.4% worst case, and that combination deserves a moment of respect: an allocation that starts below roughly $30,000 will dip under its own fee-coverage line if the worst documented drawdown actually happens, because 29.4% off $30,000 leaves about $21,200. Volatility Target Managed Rotation lists $14,000 against a 31.4% worst case, so staying above its floor at the trough implies a starting allocation in the low $20,000s. DCA Buy & Hold is different in kind: $8,000 initial plus $1,600 a month, because continuing contributions keep refilling the base during drawdowns — provided you actually keep contributing when it hurts.
Run the check yourself on whatever strategy you run. Take the fee-coverage number as the economic floor, apply the worst drawdown you can defend to your intended starting capital, and see where the trough lands relative to the floor. If it lands below, you have three honest options: allocate more, accept that bad years will temporarily push you into fee-drag territory, or find a cheaper structure for your account size. What you should not do is ignore the interaction and discover it in a drawdown, when your capital is smallest and your patience thinnest. And remember where these numbers came from: fee-coverage estimates derive from backtests, the same way the drawdown figures do, so a genuinely bad year can push you under any of these lines regardless of the arithmetic.
Exposure Scaling: Sizing as a Dial, Not a Constant
A static starting size is only half the framework, because risk is not constant — markets alternate between quiet regimes where a position behaves and violent regimes where the same position behaves much worse. The systems that survive bad years usually build the response to that into the rules, and the cleanest example is volatility targeting.
Volatility Target Managed Rotation targets 25% annualized volatility, running S&P 500 exposure through SPY and its 2× counterpart SSO, with a BIL sleeve of short-term Treasury bills as ballast. The design is exposure scaling made explicit: when realized volatility runs low and calm, the strategy can carry more exposure — up to leverage through SSO when conditions permit; when volatility spikes and markets get rough, it mechanically cuts exposure back toward bills. The risk budget stays fixed and the exposure is the dial that turns. That is the same instinct as sizing for the worst case, applied continuously instead of once at entry: the strategy assumes bad stretches will happen and pre-commits to shrinking into them rather than hoping they miss it.
Be clear-eyed about the limits, because your own version will share them. A 25% volatility target is not a 25% drawdown cap — volatility targeting reacts to realized volatility with a lag, and markets can gap straight through any target on the way down. The backtested worst case for this system is 31.4%, deeper than the target, which tells you how honest the caveat needs to be. What the dial buys you is not a guarantee; it is a designed, measured response instead of a panicked one. You can apply the same principle at portfolio level — predefine how your total exposure scales when volatility or drawdown measures cross thresholds — so that “what happens to my size when the bad year arrives” is a decision you made calmly in advance rather than one you improvise under loss.
The Framework, Put Together
So the order of operations is short enough to carry in your head. First, write down your tolerable dollar loss — the hole you can sit through without breaking process. Second, set an assumed worst case: the published maximum drawdown, then multiply it up, because published numbers come from backtests and bad years exceed them. Third, size the position as tolerable loss divided by that assumed worst case. Fourth, check the economics: after the assumed trough, does your capital still sit above the strategy’s fee-coverage floor, and are the contributions you promised real? Fifth, decide in advance how exposure scales when conditions deteriorate, rather than discovering your risk tolerance mid-drawdown.
None of this requires buying anything, and the build-it-yourself route is this blog’s home turf: design the rules, backtest them without fooling yourself, and let the discipline of the framework decide what you can run. But when the question is whether a curated, documented alternative exists for readers who would rather stand on someone else’s research, I point them to kairostrading.net. It publishes out-of-sample start dates for every system, complete portfolio reports with performance, holdings, signals, and trade history, and a Learn section with guides on systematic investing and flat-fee versus AUM economics. Members keep custody of their capital, execute in their own brokerage accounts, and membership is application-based, with each of the four current systems at the flat $100 monthly fee. Its own framing — “based on backtest; not a guarantee” — is displayed wherever performance is shown, which is exactly the posture a sizing framework needs from a source: the numbers, the method, and the caveats, all visible so the math is yours to run.
The bad year is not a possibility your framework should hope away; it is the scenario your framework is for. Size so you survive it, keep capital above the line where the strategy’s economics still work, and let the exposure dial do its job when volatility rises. A bad year then becomes an expense you planned for instead of an event that ends your participation.
Disclaimer: This blog is for educational and informational purposes only. Nothing here is investment advice. Past performance does not guarantee future results. Trading involves risk of loss.