I have the following strategy Instrument : crypto pairs (50 coins) short only ,Markets : spot and futures (isolated margin account margin sell (for spot) and futures on Binance) ,Frequency : ~5-10 trades a day ,Hold time : few hours to 2-3 days ,Capital allocation : I allocate ~8% of total capital per position and I will have a maximum of 10 positions open at any given time. The strategy does not use stop or take profit. We open a position and whenever there is sell signal and close whenever there is a buy signal.

I am using really low leverage of 0.5x i.e. if I short sell 100 dollars worth of a coin, I allocate $200 margin to this isolated margin account (or for the futures trade in isolated mode). This leverage allows the position to spike 160%-200% against our entry price before I get a margin call. Assuming the strategy should generate a certain amount of return in an ideal scenario when no margin call is ever hit but obviously that will never be the case since crypto coins are subject to 2x-10x spikes so margin calls are inevitable which means the frequency of margin calls will seriously impact the expectation of this strategy.

I chose this amount of leverage for now because it 'seemed reasonable' but from the analysis I have done from the backtest,

for some coins, I am still over leveraging because they are subject to 2x-5x(or even more) short term spikes(short squeezes and other reasons?) number of times in a year so no amount of margin allocation will save those positions from hitting margin call

for some other coins, I seem to be under-leveraging as these coins are generally less volatile and I am thinking I could get away by leveraging a bit more but then I increase the risk of getting liquidated so I need to find the right balance

Can I please get advice on how to determine the optimal leverage per position so I can allocate margin appropriately.



1 Answer 1


With the information you provided it is hard to suggest something specific but here are some general approaches you can take:

1. Your model has expected return as a output
If your model predicts the expected return of the investment (usually predictive models) you can use this number in Kelly Criterion to get the optimal leverage. There are few mathematical advantages of Kelly criterion such as maximizing wealth, avoiding bankruptcy by aggressively reducing position but you should also check it's practical limitations.

2. Use some historical factor to determine leverage
For example you can use drawdown from your backtest to decide what leverage suits you the most. More leverage for assets with smaller drawdown. Also, market regimes, instrument/sector volatility can also be a candidate. However, you will have to run backtest to see what works best and since it is based on historical data wouldn't guarantee same will apply in the future.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.