I tried out a few of these. Strongest correlation I found was between SPY-1 and NDX, fwiw. It feels like I shouldn’t be doing so much of this work in spreadsheets though, because of the time cost. Is this the kind of thing Quantopian is mainly used for?
I only demonstrated with a spreadsheet because I expect that’s something most of the audience would understand. But spreadsheets have trouble handling the amount of data we need to work with. As I said, I would probably be using Python myself. Anaconda comes with enough to do this stuff. Pandas/Numpy, matplotlib, scipy, statsmodels. I kind of prefer HoloViews for charting. I also use the AlphaVantage API to get data sometimes.
I have looked at Quantopian a bit, but haven’t really gotten into it yet. From the documentation, it does look like it can do the stuff I’m talking about.
Thanks gilch, I’ve got a lot to look into but I’m kind of excited to try this stuff out. Your series + some other materials I’ve been watching has convinced me that finding alpha isn’t as impossible as I thought.
Do you currently use any strategies whose edge you’ve confirmed by automated backtesting?
Hello glich! Thanks for writing this whole series. When I’ve first read it a year ago, I thought to myself, that instead of impulsively going to implement it right ahead, I’ll wait one year to hear from you about how your strategy worked for you, first.
So far, so good. Last year, I was up 27% at the peak. By the end of the year profits were closer to half of that. This year (2022) has been a loser so far, but we’re not even one month in yet, and it hasn’t been enough to wipe out my gains yet. My Forex bot is not doing well, having lost all profits so far and again lost about that amount, but that’s a relatively small part of my overall portfolio.
This is all within expectations.
I stand by my basic outline from the series, but I’ve refined details a bit and had a bit more insight, particularly about Nassim Nicholas Taleb’s barbell strategy, which has made me more willing to take small, speculative bets, and more concerned about tail risk.
A lot of this volatility is due to a substantial (~15%) speculative bet on Ether. I’m HODLing. Bitcoin has a futures ETF available in the US, so it’s only a matter of time before Ether joins, and the merge to proof-of-stake should be interesting. I expect a lot of volatility in Ether when those happen, especially if it’s around the same time.
Some of my winning plays were small bets on WSB short squeezes. I kind of don’t have time for the more active strategies due to work now, but most of my income is going to savings and basically all of my savings are invested in some kind of asset, and I have no plans to change that. I am hoping to gain financial independence so I can retire early. How early remains to be seen.
I’ve kind of co-founded a LessWrong Investing Seminar, which amounts to a private chat room for interested Rationalists, and an Etherpad documenting plans, tips, tricks, and accumulated know-how, but I have been doing most of the writing in that document. Activity level in the chat varies, but it’s still active. I might write up some of what’s been going in to that. I’ve come up with improved strategies, which look great in backtests, but it can take years for these to play out in real life.
It’s interesting that you cite last year as evidence of your trading going well, at a 13.5% gain, while the S&P 500 (SPY) total return for 2021 was 28.7%. Can you elaborate on your perspective given that the market performed so well in general?
I think it was actually closer to 17% by the end of last year, but mostly unrealized and this is a very rough estimate.
Good point though. Why not hold 100% SPY? Serious question! That is actually not an unreasonable long-term strategy.
But are you adjusting for risk? Stocks outperform most other liquid asset classes. The exception right now seems to be crypto. Why not YOLO 100% Bitcoin? Again, a serious question. I’m less confident that is a reasonable strategy, but people do it! Does that seem reckless? It kind of does to me. Bitcoin had an 82% (!) drawdown around 2018. Yet Bitcoin absolutely dwarfed returns from SPY in 2017. It’s really off the chart. 1500%. Seriously. SPY was up like 22%, which is pretty good for SPY, but it doesn’t even compare. A fluke? Well, even as recently as 2020, Bitcoin was up 290% compared to SPY’s 18%.
SPY isn’t all that conservative either. It has had a 50% drawdown before, and over its lifetime, its compound annual growth rate is only 9%. Its Sharpe ratio is 0.6. It’s not hard to do better than that in a backtest.
For example, 75% SPY and 75% TLT (borrowing the −50% cash) has both higher returns and lower volatility than 100% SPY, with a Sharpe ratio of 1.0. Its worst drawdown was only 27%. To be fair, TLT is not quite as old as SPY, which had a slightly better Sharpe of 0.7 over TLT’s lifetime. The 75⁄75 portfolio clearly would have been a better choice overall. But some years it underperformed SPY by a significant margin. 2021 happened to be one of those years.
Hey, I really appreciated this series, particularly in that it introduced me to the fact that leveraged etfs (1) exist and (2) can function well as a fixed proportion of overall holdings over long periods.
Is the lesswrong investing seminar still around/open to new participants, by any chance? I’ve been doing lots of research on this topic (though more for long-term than short-term strategies) and am curious about how deep the unconventional investing rabbit hole goes.
I tried out a few of these. Strongest correlation I found was between SPY-1 and NDX, fwiw. It feels like I shouldn’t be doing so much of this work in spreadsheets though, because of the time cost. Is this the kind of thing Quantopian is mainly used for?
I only demonstrated with a spreadsheet because I expect that’s something most of the audience would understand. But spreadsheets have trouble handling the amount of data we need to work with. As I said, I would probably be using Python myself. Anaconda comes with enough to do this stuff. Pandas/Numpy, matplotlib, scipy, statsmodels. I kind of prefer HoloViews for charting. I also use the AlphaVantage API to get data sometimes.
I have looked at Quantopian a bit, but haven’t really gotten into it yet. From the documentation, it does look like it can do the stuff I’m talking about.
Thanks gilch, I’ve got a lot to look into but I’m kind of excited to try this stuff out. Your series + some other materials I’ve been watching has convinced me that finding alpha isn’t as impossible as I thought.
Do you currently use any strategies whose edge you’ve confirmed by automated backtesting?
Yes, I’m trading several of those right now, including a couple that I’ve hinted at in the sequence so far.
Hello glich! Thanks for writing this whole series. When I’ve first read it a year ago, I thought to myself, that instead of impulsively going to implement it right ahead, I’ll wait one year to hear from you about how your strategy worked for you, first.
So.. How are you doing?
So far, so good. Last year, I was up 27% at the peak. By the end of the year profits were closer to half of that. This year (2022) has been a loser so far, but we’re not even one month in yet, and it hasn’t been enough to wipe out my gains yet. My Forex bot is not doing well, having lost all profits so far and again lost about that amount, but that’s a relatively small part of my overall portfolio.
This is all within expectations.
I stand by my basic outline from the series, but I’ve refined details a bit and had a bit more insight, particularly about Nassim Nicholas Taleb’s barbell strategy, which has made me more willing to take small, speculative bets, and more concerned about tail risk.
A lot of this volatility is due to a substantial (~15%) speculative bet on Ether. I’m HODLing. Bitcoin has a futures ETF available in the US, so it’s only a matter of time before Ether joins, and the merge to proof-of-stake should be interesting. I expect a lot of volatility in Ether when those happen, especially if it’s around the same time.
Some of my winning plays were small bets on WSB short squeezes. I kind of don’t have time for the more active strategies due to work now, but most of my income is going to savings and basically all of my savings are invested in some kind of asset, and I have no plans to change that. I am hoping to gain financial independence so I can retire early. How early remains to be seen.
I’ve kind of co-founded a LessWrong Investing Seminar, which amounts to a private chat room for interested Rationalists, and an Etherpad documenting plans, tips, tricks, and accumulated know-how, but I have been doing most of the writing in that document. Activity level in the chat varies, but it’s still active. I might write up some of what’s been going in to that. I’ve come up with improved strategies, which look great in backtests, but it can take years for these to play out in real life.
It’s interesting that you cite last year as evidence of your trading going well, at a 13.5% gain, while the S&P 500 (SPY) total return for 2021 was 28.7%. Can you elaborate on your perspective given that the market performed so well in general?
I think it was actually closer to 17% by the end of last year, but mostly unrealized and this is a very rough estimate.
Good point though. Why not hold 100% SPY? Serious question! That is actually not an unreasonable long-term strategy.
But are you adjusting for risk? Stocks outperform most other liquid asset classes. The exception right now seems to be crypto. Why not YOLO 100% Bitcoin? Again, a serious question. I’m less confident that is a reasonable strategy, but people do it! Does that seem reckless? It kind of does to me. Bitcoin had an 82% (!) drawdown around 2018. Yet Bitcoin absolutely dwarfed returns from SPY in 2017. It’s really off the chart. 1500%. Seriously. SPY was up like 22%, which is pretty good for SPY, but it doesn’t even compare. A fluke? Well, even as recently as 2020, Bitcoin was up 290% compared to SPY’s 18%.
SPY isn’t all that conservative either. It has had a 50% drawdown before, and over its lifetime, its compound annual growth rate is only 9%. Its Sharpe ratio is 0.6. It’s not hard to do better than that in a backtest.
For example, 75% SPY and 75% TLT (borrowing the −50% cash) has both higher returns and lower volatility than 100% SPY, with a Sharpe ratio of 1.0. Its worst drawdown was only 27%. To be fair, TLT is not quite as old as SPY, which had a slightly better Sharpe of 0.7 over TLT’s lifetime. The 75⁄75 portfolio clearly would have been a better choice overall. But some years it underperformed SPY by a significant margin. 2021 happened to be one of those years.
My portfolio is performing within expectations.
Hey, I really appreciated this series, particularly in that it introduced me to the fact that leveraged etfs (1) exist and (2) can function well as a fixed proportion of overall holdings over long periods.
Is the lesswrong investing seminar still around/open to new participants, by any chance? I’ve been doing lots of research on this topic (though more for long-term than short-term strategies) and am curious about how deep the unconventional investing rabbit hole goes.
Thank you for sharing!