What's the safe order size limit to avoid market impact

mervyn · Jun 12, 2025 at 12:05 AM

xiaosai said:
You don't understand, this is a way to estimate your strategy capacity.

Most people say their strategy's capacity is $1M, $10M, or $1B, but I think this way of describing it is actually very vague.

For example, if someone says their strategy's capacity is $10M, does that mean exceeding $10M—say, $11M—would reduce my returns, or would the returns turn negative? These are completely different concepts.

If my strategy yields a 200% return at $10M and only drops to 10% (less than the S&P 500 index) when reaching $100M, then the correct statement shouldn’t be "my strategy’s capacity is $10M." Instead, it should be that $10M is the inflection point for the strategy’s returns, after which the returns decline—but that doesn’t mean the strategy fails. It only fails beyond $100M.

If a strategy already accounts for limit order execution failures and slippage (which always happen), then the real factor affecting the strategy is the market impact of the capital.

what I am talking above is not a "what makes me comfortable, it is subjective matters" question. it is a an important step to evaluate a strategy.

If a person has experience programming a complete trading strategy independently (Not just signals, including fund management, position management, order execution, iteration and optimization methods), he will know what I am talking about.

I think most people don't care this question because they are still trading according to "feeling".
More...

silly question. if you are trading triple digit millions or more per symbol, there is a prime desk counterparty you would trade with, the trades don’t directly go into the market.

so let you know, schwab caps you at $15 million per trade (multiple of multiple executions not shown in level 2).

and if you have that much money, you hire a trader or traders to manage to execute your trade. retail shops won’t take you as a client.

xiaosai · Jun 12, 2025 at 1:34 AM

mervyn said:
silly question. if you are trading triple digit millions or more per symbol, there is a prime desk counterparty you would trade with, the trades don’t directly go into the market.

so let you know, schwab caps you at $15 million per trade (multiple of multiple executions not shown in level 2).

and if you have that much money, you hire a trader or traders to manage to execute your trade. retail shops won’t take you as a client.
More...

Maybe there is a fool who didn't understand the question himself, who said "I will trade money at 15m per trade" ?

LionsWarthogsMillions · Jun 12, 2025 at 1:46 AM

Just ask your broker directly instead of wasting your time talking to random fools on a trading forum.

But don't waste their time if you're just fantasizing, and this situation doesn't actually apply to you, yet

xiaosai · Jun 12, 2025 at 2:07 AM

spy said:
Exactly what is "a way to estimate your strategy capacity"? Can you show us?

Don't worry, it's rhetorical. My point is the people who answered 0 are just as correct as the ones who suggested there's some known fixed curve.

Of course it is, just like your question.

You asked what's "safe" and now are suggesting there's some point where "capacity is exceeded"... but you neglect that real numbers are not countably infinite.

You can do some math to approximate how much the capacity tails off, sure... but no one can tell you what you're comfortable with. Obviously you don't want to go past the point where the strategy's risk adjusted rewards go negative. But, only you can determine where you want to be when that number ∈ ℝ>0.

What's obvious to one person isn't obvious to others. Again, you're asking a question that is plagued w/ subjectivity and imprecision. So... you won't get an answer that satisfies you.
More...

Actually I don't know what do you mean "comfortable". ok, lets assume my strategy can get 200% at 10m$ level, but when it reach 100m$, it will drop to 15%, assume 15% is
my bottom line, does that mean "my comfortable number is 100m$ and 15%?"

Again, when I talk "Safe", let me explain more clearly, treat the order size as a variable X, when X <=20 contracts (that means 1-20), the strategy has same yield change speed. when X>20, there will be a significant price bias between the expected price and the actual transaction price, so I can say <20 is safe for the strategy 100% performance. is that clear?

And as you say, we of course can build math model if we have enough data, then further decide whether to close the position immediately and accept the risk of market shock, or to spread the position and accept the risk of extended closing time.

But at first, we should have data, right? of course I can do live test (and must do), throw some money to market and see what happen. but even if I do that, I can only get part of the truth. only by obtaining different data and experiences from different people can get closer to the truth. that's why I ask here.

spy · Jun 12, 2025 at 2:22 AM

xiaosai said:
Actually I don't know what do you mean "comfortable". ok, lets assume my strategy can get 200% at 10m$ level, but when it reach 100m$, it will drop to 15%, assume 15% is
my bottom line, does that mean "my comfortable number is 100m$ and 15%?"

Again, when I talk "Safe", let me explain more clearly, treat the order size as a variable X, when X <=20 contracts (that means 1-20), the strategy has same yield change speed. when X>20, there will be a significant price bias between the expected price and the actual transaction price, so I can say <20 is safe for the strategy 100% performance. is that clear?

And as you say, we of course can build math model if we have enough data, then further decide whether to close the position immediately and accept the risk of market shock, or to spread the position and accept the risk of extended closing time.

But at first, we should have data, right? of course I can do live test (and must do), throw some money to market and see what happen. but even if I do that, I can only get part of the truth. only by obtaining different data and experiences from different people can get closer to the truth. that's why I ask here.
More...

Congratulations, you're learning that reward requires risk and quantitative tools have limits. You'll do just fine.

boqinya · Jun 13, 2025 at 8:40 AM

> When calculating safe order size, beyond conventional order book depth, the forex market requires special attention to:
>
> **1. Central bank rate discontinuities**
> - For example, when the SNB removed the exchange rate floor in 2015, CHF/USD market depth plummeted by approximately 83%
> - *Solution*: Need to monitor policy risk premium in real-time using `SNB_SIGHT_DEPOSIT_RATE`
>
> **2. Cross-currency contagion**
> - Large EUR/USD orders can affect CHF/USD liquidity through EUR/CHF transmission
>
> **3. Holiday data distortion**
> - Linear interpolation underestimates post-holiday opening impact (empirical tests show CHF/USD market impact was underestimated by 35% after New Year's)
>
> How does everyone adjust forex order sizes for FOMC/SNB decision days?

SunTrader · Jun 13, 2025 at 10:05 AM

boqinya said:
> When calculating safe order size, beyond conventional order book depth, the forex market requires special attention to:
>
> **1. Central bank rate discontinuities**
> - For example, when the SNB removed the exchange rate floor in 2015, CHF/USD market depth plummeted by approximately 83%
> - *Solution*: Need to monitor policy risk premium in real-time using `SNB_SIGHT_DEPOSIT_RATE`
>
> **2. Cross-currency contagion**
> - Large EUR/USD orders can affect CHF/USD liquidity through EUR/CHF transmission
>
> **3. Holiday data distortion**
> - Linear interpolation underestimates post-holiday opening impact (empirical tests show CHF/USD market impact was underestimated by 35% after New Year's)
>
> How does everyone adjust forex order sizes for FOMC/SNB decision days?
More...

Who's talking about forex?

Besides yourself.

Forex doesn't do contracts, other than Currency Futures.