I've been swing trading the ES using a mechanical trading model, and am now broadening my reach to the electronic version of the US, CL, and GC contracts. In the ES world, the trading day ends and the settlement price is released shortly thereafter, and it's usually pretty close to the last trade, unless it's the end of the month.
But with these other contracts, it looks like settlement prices occur at various times that are basically when pit trading hours end. So if you're trading the electronic contract, the settlement price for the day will come before the last electronic trade is made. It seems to me that this is a problem if you are using OHLC data to do backtesting. Now your C is no longer your "close", but is some fixed time that could be several hours before electronic trading ends.
So that brings up several questions: 1) why is this done this way? I'm guessing it's because the settlement price was always set right after the pit trading closed, and that practice has never changed. Or is there some other reason?
2) if you want to trade the equivalent of market-on-close, (which I do), how do you deal with this? Do you build your models to treat the settlement price as the close price, and make your trades at the settlement time, instead of at the close? That might work, but it seems pretty weird. You'd have to treat the remaining part of the (electronic) trading day as the start of the next day, which is confusing. And if the low or high occurred after settlement, you wouldn't have that data to feed your trading model to calculate the next day's trades, because you would have already entered those trades at settlement time.
3) are there data vendors that provide actual close data (last trade of the day) for these contracts, and don't use the settlement data? I've been using Pinnacle, and they use the settlement price. If I could find a historical data provider with the actual close prices, that might solve my problem.
4) for those of you out there already doing this, how do you handle this? I kinda have a feeling I'm missing something here.
But with these other contracts, it looks like settlement prices occur at various times that are basically when pit trading hours end. So if you're trading the electronic contract, the settlement price for the day will come before the last electronic trade is made. It seems to me that this is a problem if you are using OHLC data to do backtesting. Now your C is no longer your "close", but is some fixed time that could be several hours before electronic trading ends.
So that brings up several questions: 1) why is this done this way? I'm guessing it's because the settlement price was always set right after the pit trading closed, and that practice has never changed. Or is there some other reason?
2) if you want to trade the equivalent of market-on-close, (which I do), how do you deal with this? Do you build your models to treat the settlement price as the close price, and make your trades at the settlement time, instead of at the close? That might work, but it seems pretty weird. You'd have to treat the remaining part of the (electronic) trading day as the start of the next day, which is confusing. And if the low or high occurred after settlement, you wouldn't have that data to feed your trading model to calculate the next day's trades, because you would have already entered those trades at settlement time.
3) are there data vendors that provide actual close data (last trade of the day) for these contracts, and don't use the settlement data? I've been using Pinnacle, and they use the settlement price. If I could find a historical data provider with the actual close prices, that might solve my problem.
4) for those of you out there already doing this, how do you handle this? I kinda have a feeling I'm missing something here.
