Most traders treat position sizing as a fixed layer beneath their strategy — a formula for calculating how many contracts to trade, disconnected from what markets are actually doing. Tomas Nesnidal runs a private hedge fund specialising entirely in breakout strategies, and he found a different way to think about it: position sizing as a live response to market conditions, not just account equity.
The result is Dynamic Position Sizing (DPS), a proprietary technique that Nesnidal credits as a major driver of his fund’s performance. The core idea is to assess the probability that the next trade will succeed before it happens, and then scale position size up or down accordingly. Not by monitoring the equity curve or win rate history, but by reading market conditions directly.
Watch the full episode below, then read on for the complete breakdown.
The problem with treating win rate as a fixed number
Traders get a backtest report and read off the win percentage. It says 51%. They treat that as a property of the strategy. What Nesnidal found is that this number is deeply misleading, because it fluctuates constantly. Break any backtest history into ten-trade windows and you’ll see win rates of 30%, then 70%, then 45%, all averaging out to the headline figure.
In TradeStation you can see this directly in the year-by-year breakdown. One year 65%. The next year 35%. The average is still 50%, but sitting in a 35% year and wondering why you’re not making money is a very different experience from reading a report number.
This fluctuation isn’t random noise. Nesnidal’s argument is that it’s driven by market conditions. Markets move through different regimes, and in some regimes your breakout strategies win far more often. In others, conditions are hostile and you win far less. If you can identify which regime you’re in before you trade, you can size positions accordingly.
What performance qualifiers actually are
A performance qualifier is a method that assesses current market conditions before a trade is placed, and outputs a probability level — not a binary signal. This is the key distinction from a filter.
A filter is binary: trade or don’t trade. It either fires or it doesn’t. A performance qualifier has multiple states. In Nesnidal’s system it produces three levels: low probability, average probability, high probability. Based on which level the market qualifies as, you scale your position size — half the normal size at level 1, normal size at level 2, double at level 3.
The qualifier is not looking at your win rate history. It’s not tracking your equity curve. It’s reading market structure directly: volatility, expansion/contraction patterns, market strength, internals, time-based patterns. The question it answers is: given what the market is doing right now, what probability level does my strategy’s next trade fall into?
The five groups of qualifiers
Through years of testing across many strategies and markets, Nesnidal reduced the approaches to five groups:
- Market contraction and expansion: Markets cycle between expanding ranges and contracting ranges. The win percentage for breakout strategies changes substantially depending on which phase the market is in.
- Market strength: A measure of directional strength in the current environment. Strong trending conditions deliver different probabilities than choppy conditions.
- Volatility: Absolute and relative volatility levels. Not all markets and strategies respond to volatility the same way, but it remains a powerful qualifier for many.
- Market internals: Independent market data points beyond price alone — breadth, internal flows, sentiment measures depending on the instrument.
- Time-based: Certain time windows and seasonal patterns deliver different statistical outcomes.
Not every qualifier works for every strategy or market. But in Nesnidal’s experience, every strategy can find at least one of these five groups that improves its net profit-to-drawdown ratio reliably.
DPS vs filters: why the approach is less prone to overfitting
Traders trying to improve performance often turn to filters: rules that prevent certain trades from being taken. The problem is filters are binary and they’re easy to overfit. Add enough filters and you can make any backtest look perfect while destroying live performance.
DPS works differently. Instead of blocking trades, it modulates size. The underlying strategy logic stays untouched. A layer of position sizing sits on top. This has two practical benefits. First, you preserve all the trades — you don’t remove the losing ones and change the strategy’s fundamental character. Second, because you’re working with a small number of levels and a small number of qualifier groups, the overfitting risk is substantially lower than adding entry/exit filters with many parameters.
Nesnidal is emphatic that DPS is not equity curve trading. Monitoring your equity curve and reducing size during drawdowns is a different approach he tried and found unreliable. Performance qualifiers read the external market, not your internal account history.
What the results look like in practice
The results Nesnidal sees vary widely. Sometimes the improvement in net profit-to-drawdown ratio is modest — 5 to 10%. Sometimes it’s 100% or more. In one example he showed during the episode, applying a single performance qualifier to a strategy improved net profit by 178% while increasing drawdown by only 29%. The net profit-to-drawdown ratio went from 4.3 to 9.3 — more than double.
The counterintuitive finding is that increasing position size in high-probability conditions doesn’t always increase drawdown. Because you’re also reducing size or skipping low-probability conditions, you sometimes improve both profit and drawdown simultaneously. The worst drawdowns typically happen in low-probability conditions, and if you’re sizing down during those periods, you cut the peak losses.
Making DPS work with limited capital
A common objection: “I’m only trading one contract. How does this apply to me?”
Nesnidal’s response: think about your one contract as your average, not your default. In high-probability conditions, trade two contracts. In low-probability conditions, trade zero. The average across all trades is still one contract. But you’re now expressing that average intelligently rather than mechanically.
In TradeStation, a zero-contract position means the signal fires but no trade is executed. That allows precise backtesting of the DPS approach even at very small account sizes.
Why this matters more than optimising the underlying strategy
The cleanest way to understand why Nesnidal believes in DPS so strongly: almost any improvement you try to achieve by tweaking the entry and exit logic risks overfitting the strategy to historical data. Adding filters, changing parameters, adjusting stop levels — every change is optimised against the past and may not hold in the future.
DPS is addressing a different question entirely. Rather than making the strategy better at identifying which trades to take, it makes the portfolio better at sizing trades based on conditions that have been shown, structurally, to correlate with win probability. The improvement comes from the market context layer, not from curve-fitting the trade selection logic.
Related episodes
- Intraday trading strategies: Gary Hart
- Consistent equity growth through diversification: Nick Radge
- Simple strategies, stable results: David Stendahl
Get the show notes & transcript
Want to build better systematic trading strategies? Subscribe to the Better System Trader podcast for weekly interviews with the world’s top systematic traders and quantitative researchers.
