Paper Trading a Signal Before It Costs You Anything
Forward testing with executable prices and a full cost model is the cheapest way to find out that a backtested edge was made of artefacts.
A backtest can be fooled in a dozen ways. A forward test recorded in real time against executable prices can be fooled in far fewer. It is slow, and it is the cheapest money you will ever spend.
What paper trading means here
Not "imagine I bought it". Specifically:
Record the decision at the moment it fires, with a timestamp, before you know what happens next.
Record the price you would actually get — the offer if buying, the bid if selling — not the mid. What a spread costs you.
Cap the size at the recorded book. If there were forty contracts available, you bought forty. Depth and slippage.
Apply the full cost model. Fees on both legs, spread on entry and exit, and a capital charge for the holding period. How fees eat an edge.
Record the exit the same way, at the time your rule says to exit, at the price then available.
Anything looser and you are simulating a market that does not exist.
What it catches that a backtest does not
Liquidity that was never there. A backtest fills at recorded quotes. A forward test records the quote, then records whether it was still there a moment later. The difference is often the entire edge.
Latency. The gap between a signal firing and an order being possible. In cross-venue work this is frequently the whole story. Stale quotes and timing.
Data problems in production. Feeds fail, timestamps skew, markets get renamed. A forward test runs into all of it.
Your own discipline. Whether you actually follow the rule when it says something uncomfortable.
How long to run it
Longer than you want to. The relevant unit is independent events, not days or trades. Fifty signals on ten underlying events is ten observations. Luck versus skill in small samples.
For a strategy with a small edge, you need enough events that a run of luck cannot explain the result. That is usually months, and if the strategy depends on rare conditions it is longer.
What to record
Every signal, including ones you would not have taken. Every non-signal near the threshold. The market state at each decision. The outcome. The full cost breakdown.
The near-misses matter: if the result is highly sensitive to exactly where the threshold sits, the threshold was fitted rather than found. Backtesting honestly.
Reading the result
Compare against doing nothing, and against the simplest possible alternative rule. A strategy that barely beats "follow the market price" is not a strategy.
Look at the distribution, not just the average. A positive mean driven by one outlier is a different thing from a consistent small edge. Variance and drawdown.
When it works
Size small and keep recording. A signal that survives a forward test has earned a real position, not a large one. Everything in position sizing and the limits of any measured edge still applies, and the forward test has not proved the edge will persist — only that it was not an artefact of your backtest.
Next
The risk category opens with how fees eat an edge.
Educational content only. Nothing here is financial advice, a solicitation, or an offer to trade. Prediction markets can move against you and a position can lose its entire value. Where this guide describes a difference between venues it describes a theoretical cross-market edge in observed prices, which is not a profit: fees, spreads, capital lockup, transfer time and differing settlement rules all come out of it first.
Keep reading
Data and Automation · 3 min readBuilding a Market Dashboard That Tells the TruthDesign notes for a page that shows prediction market data without overstating it: which fields to show, what to filter, and what to label loudly.
Data and Automation · 3 min readAlerting on Price Moves Without Generating NoiseAn alert that fires constantly is ignored, which makes it worse than no alert. How to define a threshold that reflects significance rather than movement.
Data and Automation · 3 min readNormalising Market Data Across VenuesTwo venues describe the same idea with different shapes. A documented intermediate representation is the difference between a comparison and a guess.
Prediction Market Basics · 4 min readHow an Event Contract Pays OutAn event contract settles at one dollar or nothing. Understanding that single fact explains pricing, profit, loss and most of the mistakes people make early on.