Wrong costs, in either direction
A round-number cost assumption can be off by several times. Under-estimating costs flatters bad strategies; over-estimating them quietly kills good ones, and you never find out.
The trap
Using a rule-of-thumb cost per trade (“call it 2 points a leg, each way”) instead of a cost model built up from the actual charges and checked against real fills.
How it fools you
The error can run either way, and both directions are expensive:
- Too low: high-turnover strategies look profitable when the costs would have eaten the whole edge.
- Too high: strategies that would have worked are rejected. This one is worse in a way, because you never find out. A rejected idea doesn’t get traded, so nothing ever tells you the rejection was wrong.
Option selling is especially exposed. The edge per trade is small, a few points, so the gap between a guessed cost and a real one can be the whole result.
A worked example
Take one leg of an index option strangle: sell at a premium of ₹100, buy back later at ₹100 (so we only see the costs). Here is what that round trip costs per unit, in index points, built up from the charge schedule:
| Component | 1 lot per order | 10 lots per order |
|---|---|---|
| Brokerage (₹20 per order, 2 orders) | 0.62 | 0.06 |
| STT (sell side) | 0.15 | 0.15 |
| Exchange transaction charges | 0.07 | 0.07 |
| SEBI fee and stamp duty | 0.00 | 0.00 |
| GST | 0.12 | 0.02 |
| Total per leg, before slippage | 0.96 | 0.31 |
Illustrative, at a ₹100 premium and a 65-unit lot, using the charge schedule current when written. Slippage comes on top and depends on the strike, the time of day and how you place orders.
Three things stand out:
- Most charges scale with premium, not with the index level. A rule of thumb that ignores premium is wrong for cheap far strikes and for expensive near ones alike.
- Flat brokerage dominates at small size. The same trade costs about three times as much per unit at one lot as at ten. A cost assumption that is right for one trader can be badly wrong for another.
- “2 points a leg, each way” is far too high. That is 4 points per leg round trip: about 4 times the real charges at one lot and 13 times at ten, so slippage would have to be enormous to close the gap.
Now suppose a strangle rule makes 2.5 points per strangle before costs. Charged 2 points a leg each way (8 points for the strangle), it shows a 5.5-point loss and goes in the bin. At ten lots per order the real charges for the strangle come to about 0.6 points, so even with a point of slippage it is profitable. Same rule, same data, opposite verdict.
How to spot it
- Your cost assumption is a round number nobody can trace to a source.
- The cost doesn’t change with premium, price or position size.
- Results swing from profit to loss when you nudge the cost assumption slightly: the strategy lives or dies on it.
- You’ve never compared backtest fills with live or paper fills.
How we guard against it
Every result uses the itemised cost model on Methodology → Costs, calibrated to live fills where we have them, and every strategy is re-tested at twice the cost (the “survives higher costs” check on How we score strategies).
Was this useful?
More to say, or spotted a mistake? Write to [email protected]. We read everything.