The most trusted number in finance is also the easiest to manufacture.
Show me a fund's track record and I'll show you, more often than not, a piece of fiction. Not always a lie - that's the trap. Usually it's a story: edited, survivor-selected, measured before the costs, and starting on a conveniently good day. Every number in it can be technically true while the whole thing is, taken together, a fabrication.
This is the belief this piece is here to break: a beautiful track record proves skill. It mostly doesn't. It proves the manager survived long enough to print one - which is a very different thing.
What a track record really is
When someone hands you a smooth line going up and to the right, your brain reads "competence." That instinct is exactly what the format is built to exploit. A track record is the one artifact in finance that is simultaneously the most persuasive and the least audited. People will interrogate a single fee line for an hour and accept a ten-year performance chart at a glance.
So let's interrogate the chart. Here is how the fiction gets assembled - five techniques, none of which require lying.
1. Survivorship: you only see the winners
This is the big one, and it's structural.
The funds that blew up do not send you decks. They're gone. The ones still around to show you a track record are, by definition, the ones that didn't die - so every "average manager returns X" figure you've ever seen is quietly computed over the survivors.
Run the thought experiment: put a thousand managers in a room and have them flip coins. After ten rounds, a handful will have ten heads in a row. They will have a track record. They will, in perfect sincerity, write you a letter about their process. Survivorship manufactures geniuses out of pure noise, at scale, for free.
The problem is not that the survivor is lying. The problem is that the graveyard isn't in the room.
2. Backtests: fiction with a spreadsheet
"We would have returned 240%."
Would have. The two most expensive words in asset management.
A backtest is the only place in finance where the answers are known before the questions are asked. It has more knobs than data points, so with enough parameters you can fit any curve to the past perfectly and learn nothing about the future - the model has simply memorised the answer sheet. A backtest is less a prediction of the future than a description of the past with extra confidence.
3. The conveniently chosen start date
"Since inception" sounds rigorous until you ask when inception was. Move the start date a few months and the same fund goes from spectacular to mediocre. The marketed window always begins the month things started going right.
4. Gross of fees, gross of costs
The number on the cover is almost always the number before the haircut - before fees, before slippage, before the real cost of getting in and out at size. (We have a whole piece on the gap between the quoted number and the kept number: yield is not alpha.) A gross track record is a menu price. You eat the net one.
5. The hidden drawdown
A monthly bar chart is a beautiful way to hide a path. Smooth the resolution and the worst day disappears into a slightly shorter bar. But the drawdown is the whole story - it's the number that decides whether you'd still be invested when the recovery arrived. A track record that obscures its drawdowns has hidden the only part that mattered.
Four questions that kill most track records
Tear all that out and very little is left - which is the point. You don't need to be a quant to apply the filter. Run any track record through four questions:
- Was it forward? Earned out-of-sample, after the rules were fixed - not fitted to the past.
- Was it net? After fees, slippage and the real cost of trading at size - not the menu price.
- Was it complete? Every window and every drawdown left in - not the flattering slice.
- Could it be revised? Sitting on books that can't be quietly edited after the fact.
Most records fail before question three.
What survives all four isn't really about performance at all - it's about verifiability. And a record deserves trust only when the manager also loses first (first-loss capital) and put the claim on the record before the outcome was known (the Journal).
Fine. Apply the same standard to us.
This book does not have a long, realized track record yet. The absence of one is a fact, not a marketing problem to solve - and we are not going to backfill it with simulations.
What stands in its place isn't a number. It's the machinery a number can't fake: an uneditable ledger, an honest NAV per unit, founder capital that loses first, and a dated written record that goes on the hook before outcomes are known. When the live curve exists, it will be shown whole - net, drawdowns left in.
Survivorship is the industry's default setting. Honesty is a choice - and it's visible in the plumbing, not the chart.
The one line to keep
A track record tells you who survived. It does not tell you whether survival was skill or luck.
A track record is evidence. It is not proof. The difference is where most mistakes begin.