A pool on our store publishes a maximum drawdown of −11.84%. It is an honest number, computed correctly, and it tells you almost nothing about what subscribing to that pool feels like.
Depth answers one question: how far below the previous high did equity fall. It does not answer the question every subscriber actually asks around month four, which is how long do I have to sit here. Two pools can share an identical −11.84% and differ by a year in how long they take to climb out. One of them you keep. The other you cancel in week nine, at the worst possible moment, and you never see the recovery.
Depth is a scalar, pain is an integral
The underwater plot at the top of this piece is the same equity curve you have already seen on the record page, transformed into one line: for every day, how far below the running peak was the account. It spends most of its life at zero. What matters is the shape of the excursions below it.
Depth is the single lowest point on that chart. But the experience of holding the position is the area of the region — depth multiplied by time, integrated across the whole excursion. Two pools with identical minima can have areas that differ by a factor of five.
The three numbers we publish together
Our record page will not display a drawdown depth on its own. It is always shown with two companions, because any one of them alone is misleading:
- Max drawdown % — the deepest single excursion below the running peak.
- Max drawdown duration — calendar days from that peak until equity reclaimed it. Not to the trough. To the recovery.
- Recovery factor — total net profit divided by max drawdown. How much the strategy earned per unit of worst-case pain.
The middle one is the one the industry quietly omits. It is also the only one measured in a unit a human being feels.
What a year underwater actually looks like
Below are four real shapes we see in submitted records. All four have been normalised to the same −12% maximum depth, so the only thing that differs is time.
| Shape | Max DD | Duration | Recovery factor | Held? |
|---|---|---|---|---|
| Sharp shock | −12.0% | 23 days | 4.10 | Usually |
| Staircase | −12.0% | 88 days | 2.35 | Often |
| Long grind | −12.0% | 197 days | 1.12 | Rarely |
| Failed recovery | −12.0% | Unrecovered | 0.41 | Never |
The fourth row is the one worth staring at. A pool can be sitting at an unrecovered drawdown and still advertise a max drawdown of −12%, because the number is defined on the excursion, not on whether it ever ended. On our record page an unrecovered drawdown is labelled as such, in the same red as the figure itself.
Why the arithmetic gets worse with leverage
Recovery is not symmetric with loss, and the asymmetry compounds. A 12% loss needs a 13.6% gain to get back to even. A 30% loss needs 42.9%. A 50% loss needs 100%.
Written as a rule: to recover from a drawdown of d, you need a gain of d / (1 − d). That fraction is close to linear while d is small and vertical once it is not — which is exactly why a defined-risk options pool and an unhedged leveraged one are not comparable on drawdown depth alone, even at identical depths.
The number we made ourselves publish
When we rebuilt the record engine we added a field that made several listed analysts unhappy: time to recovery, still running. If a pool is currently underwater, the record shows how many days it has been there, counting up, live, on the same page as the return.
It is not a flattering field. It is also the single most predictive thing on the page for whether a subscriber will still be there in six months, which makes it the honest thing to show. An analyst who cannot stand beside that number in public probably should not be managing your capital in private.1
None of this makes drawdown avoidable. Every strategy that earns a return has one, and a record with no drawdown at all is a record that has not been held long enough — or has been trimmed.2 The point is narrower: publish the loss beside the return, publish the duration beside the depth, and let people decide with the whole number in front of them.
Notes
- Analysts listing on the platform agree to this field at onboarding. Three withdrew their applications when it was introduced, which we consider a working filter.
- Records shorter than twelve months are marked as such on the card and cannot be sorted to the top of the store.
Arunkumar Rangpariya
Runs five pools on the platform, four of them defined-risk options structures on the weekly expiry. Writes here about risk arithmetic and how published records mislead. Has been underwater for 197 days at least once, which is where this piece came from.