Hey man, you eating venison these days? SML, do you find that the results returned from your trading activities during the half hour following major scheduled news announcements, e.g. 8:30 Thursdays, to be either more, or the same as, or less than at any other time of day?
Sure you do, "pay attention or track trading results". Christmas dinner must be pretty awesome at the SML home. best venison recipe - https://www.dogpile.com/serp?q=best+venison+recipe
I'm running an algo based on the premise that short term prices are random. And I'm putting the algo up in a trading competition for 2023 https://financial-competitions.com/ I think 1 year sample size is significant enough no? for scalping at least
Interesting backtest that price is nearly random. There is some unique behavior that es/mes exhibits https://www.elitetrader.com/et/threads/index-futures-automation.368144/page-18#post-5748094
That's not what's being discussed here, @SimpleMeLike. @hilmy83 states that because a random function resembles a price chart then a price chart is random as well.
For a day going from low to high - the day low will be found within the first 60 minutes 74 % (!) of the time on ES. Within the first 90 minutes the day low will be found 85 % (!) of the time. Sample size is the last 8 years. RTH session. ES. That doesn't sound random to me. I don't use time series for my predictive model/forecast model, but derived statistical data from time series and other stuff as well. So, before any kind of forecast is attempted there's a ton of work to be done compiling/generating various statistics and parameters. So, in the end I agree that making a forecast/prediction with a time series/OHLC data as the only input may be a futile excercise. There are some who seem to trade very well without prediction, though. There's more than one way to make money in the market.
Interesting stats. But I find suspicious that in 8 years only one time the low of the day was in the last hour of trading. It may be the case, but it doesn't sound right. I wouldn't bet 2$ on it. How come you only have 1065 data points for 8 years? Shouldn't be more like 2000?
Well, to clarify, I did say that these were statistics for days going from low to high by the close. You can call them low-high days for convenience. The other type of day would be a high-low day. Going from high to low. These two type of days will divide your data-set roughly in half. So, including all days I have 2007 trading days over the last 8 years. If you look at last week on ES, Monday was a high-low day. Next, 4 consecutive low-high days to finish the week.
The one low-high day putting in both the low and high on the last hour was the 6th of March in 2020. Now, all this knowledge may not translate to profitable trading, but knowing what's normal and what's to be expected on any given day sure is helpful.