ALGO TRADING UPDATES
- MAYA: It’s a girl! I have finally named the trading algo. And the name I came up with is “Maya“. A core tenet of Hindu culture/religion is that we are all living in a simulation (a.k.a The Matrix or call it level 2 or 3 dream state of Inception). It is all just a very vivid dream. Even Albert Einstein said once “reality is merely an illusion, albeit a very persistent one”. Well, the term we use in our culture for this is “Maya”. And what is the biggest Maya that we all know about? It is the stock market ! Hence, the name of the BOT.
- WHEN WILL THE PILOT START? As you know the algo (Maya) is live with a paper trading account and another account with real money since 10/21/2024, but since VIX is elevated, the algo will not take any trades. I can’t force it to take any trades, after all it is a rule following bot and the conditions are not right for it to take trades. My best guess is that once the outcome of the elections becomes clear (probably by Thursday or Friday), markets will take a sigh of relief and VIX should drop. Once VIX is between 16-17 range, Maya should take its first trade.Â
- ALGO BACKTEST VIDEOS THIS WEEK: I will most likely publish backtest videos for 2022 this week. I will also publish another video addressing some of our member questions about why does the BOT not take put spreads on pullbacks and corrections.
- MONTE CARLO SIMULATION (Work in progress): Although the backtests for last 10 years are extremely promising, I don’t live in a fool’s paradise and am fully aware that real life outcomes could be different. A great way to simulate the future without having a time machine is doing a Monte Carlo simulation and I have been trying to figure out how to do that. If you are aware of any good tools, please let me know. A Monte Carlo simulation will close this chapter and there is no need to accumulate any more proof that the algo works and is not overfitted to any backtests.
Elections and FOMC, Oh my!
Election results will start becoming clear within 48 hours after election day and that will start showing its effect on the markets by Thursday. We also have that pesky FOMC rate decision and the markets have priced a 98% chance of 25 basis point rate cut in November and December. Let us see what Powell does next, but these FOMC days are huge market movers, so watch out and save capital for Friday. If you are new here, I would just save most my capital for Friday and skip trading the rest of the week.
SPY – Do or Die
 SPY is at that area where not only is it a support zone around 563 but is also at the bottom of its current trading range as indicated by the lower Bollinger Bands. Since the elections are pretty much a binary event for the markets, this could result in SPY bouncing back strong. If markets slide from here, the next stop is at 543, which would represent a 4.6% pullback which is the size of pullbacks we have seen this year anyway and that does sound like a plausible outcome.
SPY – EMA crossover about to happen?
An EMA crossover hasn’t really happened, so there is no trading decision that needs to be made here, except watch for confirmation. But EMA crossovers are very reliable in predicting trend changes especially when applied to market indices and index ETFs like SPY, QQQ, DIA etc. A Trump victory could very well send these markets shooting off to the moon, so nobody knows what happens next.
DIVERGENCE INDICATORS SCREAMING “Cuidado!”Â
I use TNX (10 year treasure yields) and $DXY (U.S. Dollar Index) as divergence indicators for my trading decisions and I use VIX as a stop and go signal. Both TNX, $DXY have been screaming “watch out!”, “caution”, “cuidado!” since October 1st. And although we didn’t get a “SHOCTOBER FEST”, but markets did end up flat for October.
VIX (i.e my stop and go signal and also used by Maya (the algo) is saying ‘don’t go bullish yet’ or just put a pause on your trades.Â
EARNINGS: Very busy, but nothing earth shattering.Â
A bunch of stocks from our watchlist are reporting earnings this week, but all the earth shakers are done so I don’t expect any major price movements due to these earnings. Do note that SMCI, ARM, QCOM will affect the semi-conductor space.Â
Trading plan for next week
- Keep it small. I will be doing the same what I did last week. i.e Take 1 or 2 trades here and there throughout the week, and go back into my shell when market direction is not clear to me.Â
- FOMC rate decision could rock the markets as usual and of course there are elections.
- Maya (the algo) is doing the same and hasn’t taken any trades since it went live on 10/21/2024Â
I am super curious to understand the answer to the question as to why this is true “A Monte Carlo simulation will close this chapter and there is no need to accumulate any more proof that the algo works and is not overfitted to any backtests.” Can you help explain how/why a Monte Carlo simulation could provide such proof?
Carey, backtests do a great job telling you if you are doing something right. In my case, after running backtests, and tweaking my algorithm for 4 months, I have arrived at a sweet spot.
However, there is a well known phenomenon called “curve fitting” or “overfitting”. Basically, what that means is that you tweak your algo so that the backtests show great results, but when you go live, your algo flops. This can happen if you over tweaked your algo to perform well for the past.
What Monte Carlo simulation does is, it takes your backtests and creates hundreds and thousands of random data points and trades and creates “fake trading years” for you. When you have hundreds and hundreds of years with random stock market data, and your algo still performs well, that is sufficient proof that it will do good when it goes live. Basically, it is the best way of peeking into the future without having a “time machine”.
This is what I am attempting to achieve. Even though I am a programmer, this is quite a bit to take in. You may find this article helpful: Click Link