Just started experimenting with TRADE this week; huge leap forward in my opinion btw - will save me tons of time. I do have a question on how the slippage calculations are handled though.
I was expecting that when the TRADE transactions are filled in for a portfolio that it would automatically fill in the slippage as well based on whether the port is defined as using next open / avg hi lo / next close. This didn’t seem to happen for the couple of transactions I had though.
To me this is pretty important since it tell you how closely the actual slippage compares to the simulated backtests. Is this something that can be done with TRADE, or will we be forced to continue to manually calculate and enter those slippage #s?
Trade fills in the actual price so there is no additional slippage. The overall slippage is the difference between the Port’s Monday morning recommended price and Trade’s fill price. So a Port run through Trade will be a real series of transactions where a Sim is just an approximation of what might happen.
If you start a new Port traded through Trade, you can then run your Sim starting at the same date (so it is buying the same stocks at the same time) you can then compare the running history of the Sim to the Trade Port.
Thanks Denny. The challenge is when the sim over the same period diverges from my real port I’d like to see what the slippage on the real fills was relative to the hi / lo avg that’s used in the Sim. To do this I had been recording this slippage info on my ports on each transaction manually so I could see whether the % slippage amount I used for my sims approximated what I was seeing in real life. Fairly painful manual process to be sure so I was hoping that TRADE would automate some of this for me. Don’t suppose there’s a smarter / easier way to do this?
Slippage is an artificial deduction in sims to account for market impact and bid/ask spread. We can’t deduct any slippage on real trades. This slippage is real cash and the numbers won’t add up with your brokerage statement if we create artificial deductions.
In a real trade if the order executes at the current realtime midpoint of the best bid/ask price then the “slippage” could be defined as 0. Right now we cannot record what the realtime bid/ask is at the exact moment you place your trade. What we can do is present the end-of-day open-high-low-close on the day of your trade on all completed trades, let you download all this info into excel, and you can then easily figure out if the TRADE algorithm you chose is beating the EOD hi/lo avg, or whatever other EOD formula.
Since Trade prices are real and Sims are Simulations, it seems to me it would be a lot easier to adjust the slippage in the Sim until the performance matches your real account from the Trade prices.
You may need to start the Sim on the same date that you started trading through Trade. This may take awhile to have enough Trade results to make the Sim performance’s estimate of slippage converge on the real performance from Trade. It would take a similar amount of time to download enough data for Excel to calculate an accurate slippage.
To me, the only useful value of slippage to use in a Port is the total slippage between the Port’s recommended previous close price and the real fill price. That is the value I run in my Ports, which is higher than the value I run in the Port’s Sim (which uses next open prices). To get the delta slippage I add to the Port’s slippage, I calculate it from the difference of 2 Sims, one run with previous close, and one with next open.
Yes this could work assuming the simulation and the port had exactly the same transactions, but in practice I find there are usually differences due to various other factors (missing the occasional rebalance in the port, vacations, the odd override on recommendations for whatever reason), and that makes it harder to tease out the impact of actual trade execution by itself. Having the end-of-day prices available in the excel download would make it possible to look at exactly how much money was lost due to simulated vs. real execution prices and I think make it easier to identify differences from simulated performance related to slippage vs. other factors like deviations from the simulated txns.
Marco I’m hoping including the EOD prices in the download excel would not be difficult (?) If that is the case any thoughts on when we might be able to get that support?
