So many have asked for tutorials on some quant strategies. So this is my first tutorial for some basic quant trading strategies. This is not really a strategy in and of itself, this is to help you determine realistic price points as part of your overall strategy.
You will need Excel to do this.
If you like this kind of tutorial/find it helpful, let me know and I can continue posting similar stuff on how to apply some more basic quant strategies into your trading.
Take care and trade safe!
Comment
⋅
Just wanted to point out some things I skipped over in the video because I either forgot to mention or I didn't really think about (I live and breathe statistics on a daily basis and sometimes I just think people also have statistics backgrounds haha, I try to be very easily understood but I am not always unfortunately):
Data Updating: For probability assessments, probability is in constant flux. Every passing day, probability shifts and thus its important to keep your probability models up to date with recent data! I personally update mine weekly, however during 2021 , I would do it more frequently (But to be honest, I didn't rely a lot on probability in 2021 because the market had one direction, up. I have relied on these assessments 100% more this year).
Why/How it works Prospectively So, for prospective assessments, which is the application I intend to use it for here and how I have displayed it to be used, the ideal situation is for the data to be as linear as possible and as normally distributed as possible.
The best way to determine this is by running statistical analyses and histograms; however, this requires a bit more of a higher level understanding of stats concepts. To simplify this, the reason I explained how you should assess your time frame the way I have shown is because this ensures you are getting the most, albeit subjectively, linear and normally distributed data you can get without doing complex stats calculations and tests. Out of curiosity I ran a test on the time frame chosen for this video, and it was actually pretty on point with how the distribution turned out! Meaning the data is actually reliable in its power to prospectively ascertain future events (kurtosis was slightly less than 0, meaning slightly skewed but VERY close to normal distribution).
If you deviate from this method of determine your data, you may take away the predictive power of the probability assessment.
Perhaps I will do that as my next tutorial, show you how to check distribution, etc. and what you can use this for to make predictions.
But the key take away is if you follow the steps in this video identically, the data should be reliable enough to draw prospective assessments of probability! But just remember to update it and frequently.
Thanks for watching/reading and leave your questions below!
Great post. Math probabilities are what built Vegas. Everything is math. If you have a 55% chance probability of success you should put 15% of your investment into it. Yes, you can lose. But over a longer timeline you will win. Play the probability and trends.
stein3d
⋅
Awesome tutorial, thanks for posting!! So with probabilities given that you're looking at a certain timeframe of historical data, does that project forward as well? For example you looked at data back to April 2020, ~2 years, would those percentages apply to the next 2 years? Does that make sense? Really just trying to get a better understanding of the criteria for determining an appropriate historical timeframe, and whether that has any bearing on the timeframe for future targets to potentially be realized.
Steversteves
⋅
@stein3d, Great question! I should have covered this in the video and totally forgot, I was just trying to keep it short. Yes, so this type of analysis favors more towards bear markets than bull markets. It does work prospectively in bull markets, but the data requires constant updating frequently (like every few days vs every week in a bear market). But the probability is in constant flux because the stock is in constant flux, so as the stock consistently closes lower and lower, the probability shifts more to the negative side and vice versa. For me personally, I will use this to supplemental time series modelling and calculations a month in advance. I had a price target on BA for 147 last week based on time series modelling I did over a month ago, and I checked the probability level that it was actually realistic (and it was! with a probability approaching 50%). I hope this answers your question but let me know if it doesn't or you want more clarification!
stein3d
⋅
@Steversteves, Yea that clears things up pretty well, thanks! Definitely makes sense that the data driving the probabilities would need to be updated regularly. I'm still trying to wrap my head a bit around how you know the range of data you're using is sufficient to calculate a probability (why not use the 1 hour close prices for the last several months for example), but seems like I should probably do some more digging into probabilities to educate myself.
Steversteves
⋅
@stein3d, Ah! okay so that is a step I skipped in this video because it relies a lot on theoretical statistics. But you can achieve it sort of similarly by using TradingView's regression tool. When I put in the data, I don't actually use excel, I use SPSS and SAS, but I will do a curve fit on the data and a histogram to determine the distribution of the data (whether its equally representative and as close to normally distributed as can be expected in real world applications). I will make adjustments based on that and draw probability calculations off of that data once it meets those requirements. Its hard for me to teach that on TV because there is a lot of math theory understanding. However, to sort of correct for that, is why I explained how you should determine your date the way I explained it (drawing a trend line and making sure there is equal representation above and below, this will likely lead to more linear and more evenly distributed data). If you use TV's regression tool, you can look at the R value it prints and the closer to 1 it is, the more of a better "fit" the data is, i.e. more linear and better results.
Tradersweekly
⋅
Thank you!
SteelTrade
⋅
Very interesting. Thank you for sharing this with us. I am curious how you would continue to monitor this after entering a trade. If you have a target of 313 in a swing trade, would you continue to update the data and reassess your position? If the market moves against you, and the probability lowers significantly that your target will be hit, would you choose to exit the trade?