if the model is in a drawdown and some conditions are met it will buy or sell better up to 3 times, usually the entries are clean so it's rare but it has that option. each add on entry is harder and harder to meet criteria so if the first add was better by .50 the second would be by 1.0 and the 3rd by 2.0 points and a 4th by 4.0 points but that never gets hit cause the stop is there.
Reversion to the mean only trading...
- Thread starter MarkBrown
- Start date
if the model is in a drawdown and some conditions are met it will buy or sell better up to 3 times, usually the entries are clean so it's rare but it has that option. each add on entry is harder and harder to meet criteria so if the first add was better by .50 the second would be by 1.0 and the 3rd by 2.0 points and a 4th by 4.0 points but that never gets hit cause the stop is there.
Do you trade ES only? How many trades do you average per day? What is the average timeframe per trade?
Profit target is exiting on a limit and it seems to be entering on a limit as well. The execution assumption seems to be 100% filled on touch, which is the best case. This inflates the backtest.
1,000,000 in 100 days with 10 contracts is 1,000 per day per contract. With 80 trades a day it makes a tick per trade. Remove a tick due to backtest assumptions and it's even.
1,000,000 in 100 days with 10 contracts is 1,000 per day per contract. With 80 trades a day it makes a tick per trade. Remove a tick due to backtest assumptions and it's even.
How do we subscribe, I have spare cash right now
The law of large numbers and either conservative capital allocation or an algorithm that detects higher risk situations or utilization of consistently negative correlated assets should create very predictable and reliable automated system.
Actually, I'd say it sounds great to the typical, new trader.
With an actual mql4 code and accurate backtesting, it might be reality.
Jim Simons is a real person. He really makes lots of money with statistical analysis. So it is possible.
i was actually the very first person to expose the inaccuracy of using limit orders in backtesting as well as bouncing ticks when backtesting and the replacement of the closing price of the day with the settlement price which cost me 300k from a friday close in bonds to the monday open when a model that was short on friday was still long on monday after recalculating for the settlement price as opposed to the close price.
the model does use limits on entry but those entries are only on strong momentum in opposing direction which provides at least the price or better often as i stated earlier in the thread. the exits are at market when the profit target has been met, nothing as fancy as the entries. sometimes their better fills and sometimes their worst.
even though the model trades 24hrs i only use it live from 0859 to 1240 chicago time, if i traded it any other time the expected positive slippage on entries would be far less likely. but it over performs during the before mentioned time slot.
thank you for your comments
He didn't say anything of substance, but if you can translate what he said into a strategy, test it out, and let us know.
Here's what I heard in everyday English:
This:
"The law of large numbers and either conservative capital allocation or an algorithm that detects higher risk situations or utilization of consistently negative correlated assets should create very predictable and reliable automated system."
Is just a fancy way of saying this:
Over many trades, by risking a little at a time, and avoiding high risk situations, and diversifying; you'll have a great system.
I'm surprised he didn't add some fancified rendition of "buy low; sell high."