Using OHLC Data to Model Sell Prices in Backtests

# How to estimate probable seeling pricegiven OHLC data for backtesting?


# How to estimate probable seeling pricegiven OHLC data for backtesting?












I'm relative new to this, so I might be asking something that doesn't make sense. Here is my scenario:

I have intraday day at 1 minute intervals. This data has ohlc data and I want to compute for any given interval what the likely sell price would be. I could just assume the worst case and take the low price, but I'm assuming there is something a little more accurate than that.

I get that there is no way to accurately predict what the sell price would be, since an actual order potentially changes the outcome. I just want to know if there is a best practice for predicting what the sell price would be if I tried to execute an order on a given interval using historical data.

## Answer by d0rmLife (score 0, accepted)

https://quant.stackexchange.com/a/21204

It sounds like you are trying to backtest a strategy and want to simulate actual trading conditions.

If this is the case, I make the suggestion that liquidity concerns are of more relevance than nailing a hypothetical order price in a 1-minute interval. If you're really concerned about the price, you could simply average the prices. Note that this is mathematically equivalent to weighing each value at 25%.

You could get more industrious and estimate the variance within the interval and make a guess about what that says about the hypothetical price, but you would need to verify that assumption empirically, anyways. To me that sounds like a lot of unnecessary work.

When you live-test a strategy there are many variables that are present that are not easily modeled in a backtest. So to really work at resolving just one (that is minor IMO) is perhaps not the best use of your energy.

## Answer by Shahar (score 0)

https://quant.stackexchange.com/a/21180

It probably depends on the liquidity of your asset and the size of the sell order: if you want to trade a heavily traded [and thus liquid] ETF such as SPY for example, your sell order would probably fill almost instantly - unless it is a very big order. In that case, obtaining second or tick data would remove most uncertainty from the question.

However, if the asset is not liquid and/or a very large order is sent, the sell would certainly be filled around the low [and yes, possibly lower].

Finally, lacking any additional information, I would either use the Close price, which usually is representative of the time period, or the low, as a [proxy] worst-case scenario [keeping in mind that it is not exactly worst-case].