Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Hi All... am in the process of backtesting a futures system, but am unsure of how large the one-way transaction costs I should assume for the following futures:
First question: Is this a fixed system, already designed, and you want to know how to allow for commissions or is this a system that is in the process of being designed by an artificial intelligence or data-mining process and the commission rate changes the decision making process? If it's a system that is already designed, try $15 per side and that should cover commission and slippage if you are paying around $2 per side in actual commission. If you pay more, increase the commission rate above $15 per side.
If this is a system being produced by artificial intelligence/data-mining you must ask yourself "What do I want the system to do by way of trades per day or days per trade?" Presuming that you are happy to accept whatever the system generates (this might not actually work for you in practice as it may not suit your trading philosophy), then initially I would backtest using zero commissions and check the results. If the trade numbers are low and the profits are low (or worse) then you can probably put this trading system into the "highly likely to fail category". If the profits are low and the trades are numerous, try gradually raising the commission rates to slow down the trading rate. Does that improve the profits? Different commission rates will change the rules of the system.
I will leave the answer there for the time being as this can get quite complex, depending on how the trading system is being developed.
Thanks OldGrey... this is for a system that has already been developed.. thanks for the figures.. I assume it's relatively robust to assume similar transaction costs across different markets, e.g. FX, Energy, etc?
My experience is that there is normally only a slight difference, if any, in transaction costs across different exchanges but it may well depend on your broker. It also depends on whether this strategy is executed electronically or manually. That can make a huge difference.
Before I even checked this I would make sure the basic system is robust, checking on the type of data used to create it, the comparison with current markets, the blind test results and the walk forward results. This is where your real risk is sitting.
i would never use any transaction or slippage cost when backtesting. doesn't mean that i ignore the cost it just means that have good reason to deny it's presence in the testing process.
What is your good reason to "deny its presence" and leave it out of a backtest? Transaction costs and slippage/spreads are real.
Obviously if you don't take many trades then maybe it is trivial to factor such costs, but the more trades you take the more you need to be on top of the actual transaction costs.
let me spark your interest in my answer with a question: in what case would i not only "not" include cost in backtesting to be realistic but yet would in fact have to add profits to the backtesting to make it realistic?
if you have transaction and slippage cost which would effect system performance backtesting then i would explore ways to negate that cost somehow, turn a negative into a positive.
i did state "prior" that i did not ignore cost but rather i just did not include it in the backtesting of models.
Im confused. You have to pay to get into and out of every single trade. No exceptions (that im aware of?). Therefore by not taking this cost into account in back testing, you are removing a certainty that exists in live trading from your testing. The more often your system trades, the less accurate the back test results will be.
some people have credit cards that pay points and some do not. even when you get points you paid for them right so what did the points cost that you didn't receive but yet you paid for them.
mark brown
“To believe a thing impossible is to make it so.” - French Proverb