I think everyone uses some sort of mental checklist (signals) and the more of those that align the more confidence/stake/liability you commit to. But everyone probably has their own unique blend of herbs and spices behind their decisions.wearthefoxhat wrote: ↑Tue Mar 16, 2021 12:32 pmI believe PW also uses signals to confirm or unconfirm his decision making.
The rolling high and low baf Dallas made is pretty handy, you can store the high & lows of 30 second time frames (or whatever you like) & then look to narrow your trade time window down to 5 or 10s to take a scalp of the rolling high or low. You could even broaden further to find directional bias and look at periods starting from 5 mins to find directional bias then start narrowing the time frame down to 30 second periods for highs and lows and then down to 5s periods to take scalps of those highs and lows.goat68 wrote: ↑Mon Mar 15, 2021 10:43 pmSo Peter and co are in trades of several minutes... but yes i've realized it doesn't seem possible.ShaunWhite wrote: ↑Mon Mar 15, 2021 10:23 pm"only" a minute or soHow about starting at 1 or 2s, afterall a string of 2s predictions IS a prediction over a minute.
I'd be interested to find out why you think a 1 minute prediction is possible? Even manual traders with all their intuition and additional information would struggle do 1 minute forecasts anything like reliably, unless one of them can say otherwise.
Equally I don't really get the 1second time period either... the price hardly moves in 1 second typically, but I guess a stable scalp it doesn't have to.
You won't be able to backtest offering without a good simulator or at the very least some fill rate data.
& backtesting is like watching a porn movie in attempt to get the porn star pregnant... how do you even know you're not back testing the anal category... chuck back testing in the bin, you're wasting your life, forward not backwards.I like your attitude James!jamesg46 wrote: ↑Thu Mar 18, 2021 7:06 amScrew waiting to be matched.
Give your automation a set of gloves, chuck it in the ring (literally) and tell it to punch when there is weakness at the high or at low. Test the money sat waiting & if you've got direction & momentum on your side, weakness on the opposite side go for the uppercut, take the money thats there, maybe there is a reason for little resistance, maybe people are scared to enter the ring at that moment in time.
Make you own fill rate
I've tried to trade SMA (20s and 60s) crossings manually 2-3mins before official start-time. First few weeks result was impressive and then it slowly degraded to modest result, then I kinda got disappointed and decided to put it to the back of the drawer.goat68 wrote: ↑Wed Mar 17, 2021 11:34 pmSo being busy backtesting and a few live bot attempts to see behaviour of a scalper type bot. I took the rolling high/low idea and the idea of having a "signal" that relates to my high and low "value". The signal is a metric roughly indicating backing/laying, and my current bot basically takes the SMA(10second) and SMA(40second) of it and back's when above lays when below. But can't make it turn a profit. What I typically see in backtesting is lots of small losses due to volatility, or the indicator wobbles for very little real price moment causing more spread loss....frustrating.
Back to drawing board again...
thank you for the tipsnapshnap wrote: ↑Thu Mar 18, 2021 11:59 amI've tried to trade SMA (20s and 60s) crossings manually 2-3mins before official start-time. First few weeks result was impressive and then it slowly degraded to modest result, then I kinda got disappointed and decided to put it to the back of the drawer.goat68 wrote: ↑Wed Mar 17, 2021 11:34 pmSo being busy backtesting and a few live bot attempts to see behaviour of a scalper type bot. I took the rolling high/low idea and the idea of having a "signal" that relates to my high and low "value". The signal is a metric roughly indicating backing/laying, and my current bot basically takes the SMA(10second) and SMA(40second) of it and back's when above lays when below. But can't make it turn a profit. What I typically see in backtesting is lots of small losses due to volatility, or the indicator wobbles for very little real price moment causing more spread loss....frustrating.
Back to drawing board again...
What I remember it's that type of a race (if we are taking about UK racing) is critical.
Maybe try something with VWAP (not to be confused with VWMA) and long-period SMA -
https://www.youtube.com/watch?v=rZRt-miYr2I.
Targeted bursts, with simple questions and definative answers. Explore the influence of every metric regardless of each one alone being profitable. Then when you find a contender (small loser or a breakeven) you can feed what matters and more importantly what doesn't into the mix to refine it. Establish a methodology, apply it conscientiously and eventually you'll find something. A haphazard approach will mean you miss things and have a desk full of half finished ideas, and it will all have been a waste of time and money.goat68 wrote: ↑Thu Mar 18, 2021 3:07 pmOn a slightly different aspect. I'm going to try and spend less time in the evening staring at charts and stats, I think too long a time at once confuses my thinking.... Fresher shorter bursts I think. I should follow my own advice on Marathon training, concentrate just as much on "recovery", don't stress the body too much at a time.
If you can't see anything tradable you sit on your hands, and no single strategy is applicable to all markets anyway (well no single strategy you're likely to find soon). You're only looking to get about 55%+ right and even the best might only be making a profit on 75-80%. It's an imprescice science with outcomes expressed as probabilities rather than binaries. If that race doesn't comply with a theory you've had it really doesn't matter in isolation.
