When The P&L Is Too Good
A strategy that's suspiciously smooth or suddenly outperforms everything nearby usually isn't a breakthrough — it's a stale price, a booking error, or a mark that hasn't caught up with reality yet.
Prerequisites: Reading Your Daily P&L
Every experienced trader has the same reflex: a day's P&L comes in a lot better than it has any right to be, and the first move isn't to celebrate, it's to go find out why. Not because good days are suspicious in general — most of the time a great day is just a great day — but because the P&L numbers most likely to be wrong are the outsized good ones, and being wrong here in either direction has real consequences. Booking a phantom gain today just means you book the correction as a loss later, usually when you've already sized up on the strength of the number that wasn't real.
The usual culprits
A stale or wrong price. An illiquid position gets marked off a data feed that hasn't updated, or a corporate action (a split, a dividend) hasn't been applied, and the system computes a gain against a price that's simply wrong.
A booking error. A trade entered with the wrong quantity or the wrong side — a "buy 50,000" that should have been "buy 5,000" — inflates the position's apparent exposure and, if the stock moved favorably, its apparent P&L.
Model marks versus reality. For anything marked to a model rather than a live market price (thinly traded credit, some derivatives), a model input can be stale or simply wrong, producing a mark-to-model gain that a real transaction would never confirm.
Genuine alpha. Sometimes it really is just a very good day. The point of the check isn't suspicion for its own sake, it's ruling out the first three before you believe the fourth.
Worked example
A small-cap position shows a P&L of +$310,000 for the day on a stock that's normally quiet, against a typical daily P&L for that position closer to ±$15,000 — a 20x jump. A quick check: the position is 200,000 shares, and the system's closing price is $46.55 versus yesterday's $45.00. That's only a 3.4 percent move, which on 200,000 shares is — the arithmetic is internally consistent, so the error, if there is one, is upstream in the price itself.
Checking a second data source, the stock actually closed at $45.10, not $46.55 — the primary feed pulled an after-hours print from a thin, erroneous trade. The real P&L for the day was , about $20,000 — in line with the position's normal range. Left uncorrected, the $310,000 phantom gain would have shown up in performance reporting and possibly triggered a size-up on a stock that hadn't actually moved.
The habit worth building
Set a simple flag: any position whose daily P&L exceeds some multiple of its normal daily range (five times is a common threshold) or exceeds a fixed dollar amount gets a five-minute check before it's reported anywhere — confirm the price against a second source, confirm the position size against the fill blotter, confirm no corporate action was missed. It costs almost nothing on the days it turns out fine, and it saves you from reporting, and possibly trading against, a number that isn't real.
Outsized good news deserves the same scrutiny as outsized bad news. The cost of checking a real gain is a few minutes; the cost of not checking a fake one is a size-up followed by a much bigger correction later.
The correction from a phantom gain doesn't disappear — it shows up later as an "unexplained" loss when the price or booking finally gets fixed, which is confusing if nobody remembers that the earlier gain was never real in the first place.
Further reading
- Kissell, The Science of Algorithmic Trading and Portfolio Management (ch. 3)