MyAlgoGuru Help
Education Glossary Open the app
Chapter 3 of 5

Slippage, the cost no invoice shows

Brokerage arrives as a line item. Slippage arrives as a worse fill, and the only way to know what it costs you is to measure it yourself.

Every cost in the last two chapters appears on a contract note. You can add them up, check them, and argue about them.

Slippage never appears anywhere. It is the gap between the price you decided at and the price you got, and because nothing itemises it, most traders assume it is small. For a liquid index option at modest size it genuinely is. For a cheap far strike, or at size, or in the last ten minutes before expiry, it can be larger than every charge in the previous two chapters put together.

Three things make up the gap

The spread. Buying at the offer and selling at the bid means you pay the spread every round trip. If the bid is 119.90 and the offer is 120.10, a buy-then-sell round trip surrenders ₹0.20 per unit before the market has moved at all. This part is not bad luck, it is the price of immediacy, and it is the floor beneath which slippage cannot go.

Latency. Between your decision and the exchange's acknowledgement, time passes — tens of milliseconds for an automated order, seconds for a human one. The market does not wait. On average this is a wash, but it is not a wash on the trades that matter: the ones where price was already moving, which is exactly when your signal tends to fire.

Impact. Your own order consumes the book. The first hundred units fill at the offer; the next four hundred fill at progressively worse prices because you have eaten the offer. For retail size in index options this is usually negligible. In a stock option two strikes out of the money, where the whole visible book might be a few hundred units, it is the dominant term.

Why it is never in your favour

There is a tempting argument that slippage should average out: sometimes you get a worse fill, sometimes better.

It does not, for a structural reason. You cross the spread to get filled. A buyer who wants certainty pays the offer; a seller who wants certainty hits the bid. Both are worse than the midpoint, and both are worse by design — that is what the market maker is being paid for. Averaging applies to the latency component only, and even there it is biased, because your signals are correlated with movement.

This platform's backtester encodes exactly that asymmetry. Slippage is entered as a percentage and applied against the trader on both sides of every leg: a buy pays more and a sell receives less on the way in, and the sign flips on the way out, because closing a buy is a sell. The result is always worse than the unslipped run, which is the only honest direction for an assumption you are making about your own execution.

The detail that changes more than the number

Here is the part most cost models get wrong, and it is worth understanding because it changes which trades you take, not merely what they earn.

There are two places you can apply slippage: to the bottom line, or to the fill.

Apply it to the bottom line and every trade's result shrinks by a fixed amount. Simple, and wrong — because your stop and your target are still being measured against the clean price. The run exits at exactly the same minutes as the unslipped one, just with a smaller number. That is not what slippage does.

Apply it to the fill — which is what this platform does — and the slipped price is what the stop and the target are compared against. A position whose clean price never quite reached the stop now reaches it.

stop level entry clean run rides to target slipped run is stopped out here and never sees the move that followed The two runs are the same rules on the same day. The only difference is where the fill is assumed to be, and it changes the trade from a winner into a loser.
Slippage on the fill is a different experiment from slippage on the P&L. The clean path (blue) bottoms out just above the stop and goes on to a profit. The slipped path (red) is the same path a fraction worse, which is enough to trip the stop — so that trade exits early, at a loss, and is absent from the rest of the day. This is why a slipped backtest has a different trade count from an unslipped one, not merely a smaller profit. If your two runs have the same number of trades, slippage was applied to the wrong place.

What number to use

Do not guess. Measure.

The measurement is simple and nobody does it: for each of your own fills, record the quote at the moment the signal fired and the price you actually got. The difference, averaged over a few dozen trades, is your slippage — in your instrument, at your size, through your broker, with your latency. It is the only number that applies to you.

Until you have that, the spread gives you a defensible floor. Take the typical half-spread in your instrument, double it for a round trip, and treat it as the minimum. Then add something for latency and impact, and be more generous the less liquid the contract.

And express it as a percentage of premium, not in points, because that is where the surprise lives.

Illustrative

The same absolute spread, two different strikes.

At-the-money weekly, premium ₹120, bid 119.90 / offer 120.10. Half-spread ₹0.05. A round trip crossing both ways costs ₹0.10 per unit — 0.083% of premium. Negligible beside the 0.79% of statutory charges from chapter one.

Far out-of-the-money, premium ₹2.50, bid 2.40 / offer 2.60. Half-spread ₹0.10 — twice as many paise, on a premium one forty-eighth the size. A round trip costs ₹0.20 per unit — 8% of premium.

Eight percent. Before brokerage, before STT, before the market has moved. A strategy selling cheap far strikes for a 20% return on premium is handing a third of its gross edge to the spread, and that is in the quiet case where the book was there when it looked.

This is the real reason cheap options are dangerous, and it has nothing to do with the greeks. The premium is the denominator. A spread that is trivial on an expensive option is enormous on a cheap one.

Where slippage stops being an estimate

Three situations deserve naming, because in each one slippage stops behaving like a small tax and starts behaving like a risk.

The last minutes before expiry. Spreads widen, the book thins, and the price you see has often not traded for some time. An exit instruction given at 15:25 on expiry day is a request, not a transaction.

Size against a thin book. If your order is a meaningful fraction of the visible depth, you are not paying the spread, you are setting the next price. The cost is not the half-spread; it is however far down the book you have to reach.

A quote nobody traded at. Many option strikes do not trade for long stretches, and a last-traded price from forty minutes ago is not a price you can transact at. A cost model fed those quotes produces an answer that is arithmetically correct and practically fictional. If a strike's quote is stale, the honest response is to treat the series as unavailable rather than to slip a number that was never real — which is the rule the next module applies to every reading taken off the chain.

What to take from this chapter

  • Slippage never appears on a contract note, which is why it is the cost most often assumed away.
  • It has three parts: the spread you cross, the latency you cannot avoid, and the impact you cause yourself.
  • It does not average out. Crossing the spread is structurally against you — that is what you are paying for.
  • Apply slippage to the fill, not to the bottom line. Applied to the fill it changes which trades happen; applied to the P&L it only shrinks the number.
  • A slipped run and a clean run should have different trade counts. Same count means it was applied in the wrong place.
  • Measure your own: record the quote at signal time against the price you got. Until then, use double the half-spread as a floor.
  • Express it as a percentage of premium. The same ₹0.10 spread is 0.08% on a ₹120 option and 8% on a ₹2.50 one.
  • Near expiry, at size against a thin book, or on a stale quote, slippage stops being a tax and becomes a risk.