Strategy 0
This post simply consists of commentary. I am not a market adviser. NOTHING IN THIS POST IS INTENDED AS INVESTMENT ADVICE.
In this post, I will lay out the simplest strategy I can devise to outperform the S&P 500. Since picking stocks is anything but simple, I will not make this strategy a stock picking strategy. Instead it will be a market exit and re-entry strategy for investments kept at market weight in the S&P 500. Timing exit and re-entry is itself complicated, and not something that someone can do with a lot of confidence. So instead of looking for a way to completely exit and re-enter the market, I'll talk about reducing a position and strengthening it again.
If you're willing to go with an empirical strategy as opposed to one with more of a theoretical justification. Looking at the 100 period exponential moving average has actually been a remarkably accurate predictor of stock market collapse. Sell as soon as the market crosses below, buy when the market crosses back above. (But this strategy does have a reasonable theoretical justification. 100 periods because 100 is a nice round number -- which is evidence that I didn't tinker with the strategy in search of a perfect fit but just tested something "obvious" -- exponential moving average because averages are easy to calculate and returns are basically compound interest so using exponential as opposed to arithmetic averages makes sense. I have not actually looked at the arithmetic average to see how well in predicts anything, because it's arithmetic which means it really isn't mathematically suited to describe the market. I plan to use it, or at the very least, take it under advisement in my investing. UPDATE: I did tinker with the numbers a little bit, and you get better returns for periods in the range 30-60 than you do for a period of 100, but the smaller the number you choose, the more in and out there is, and the less significant the predictions are. Crossing in the 100 basically means the market is about to collapse, but crossing with smaller numbers doesn't indicate that nearly as strongly. If I was using a smaller number, I would sell as soon as it crosses below, but wait for it to remain above for 48 hours or so, and make sure it stays above for that whole time -- but the more detailed strategy has much more potential to be overfitting the data.)
Throughout this example, I'm going to be assuming that a person as a certain particular amount to invest, and does not have another source of income to use for adding money to the investment. I'll also be assuming that no one is deallocating money that had initially been set aside for investment. In the event that someone is additional money or removing existing money from the investment account, everything about this strategy should be re-normalized. (In other words, "selling" means reducing the percentage of money available to invest that is invested, and "buying" means increasing the percentage of money that is available to invest that is invested. If the strategy says to sell until 90% invested and keep the remaining cash on hand and someone adds another $10,000 to the account. $1000 of that should be added to the cash on hand, and $9,000 should be invested if the person is abiding by this strategy even if the strategy says not to by buying.)
The stock market is not efficient, nor is it a random walk. It exhibits certain consist patterns. In particularly, the market moves in bear-bull cycles. These cycles have a definite upward pattern. On average the highs of each bull market are higher than the highs of the previous bull market, and the lows of each bear market are also higher than the lows of the previous bear market. At the top of each bull market (or frequently even a year or two before it reaches the top of a bull market), the stock market hits record highs. However, during corrections, it does not hit record lows. In most corrections it doesn't even hit ten year lows. It has never hit a thirty year low.
From this observation alone, we have singled out a very specific market event that is worth noting, and can define a strategy that revolves around that particular event. The notable event is the market hitting new record highs.
Ideally, you want to buy stocks when the market bottoms out and sell them when it peaks, but timing this ideal is impossible. However something that is possible is to only sell when the market is at record highs, and only buy when the market is not at record highs. Of course, once the market has hit a record high, every new move upward is a new record high. Every move downward below any of those record highs are no longer record highs. Thus, it is still possible to underperform the market while only selling at record highs and only buying when the market is not at record highs. For example, the market could reach 2150. Someone could sell everything at 2160. The market could continue climbing all the way to 2500, before contracting to 2400 which would no longer be a record high. Someone could then re-enter at that point and underperform the market by 10% while abide by the two rules I've just given. So they don't define a winning strategy yet. Actually, since they don't give much definition about when to buy and when to sell, they aren't much of a strategy at all.
Actually, it's possible to be quite a bit more nuanced than this. If the market will ever hit a value below it's present value again, you benefit by selling so long as you buy back when it is at a lower level. The ideal time to buy stocks is whenever the market hits a level that is low enough that it will never reach that level again.
Bull markets and bear markets can be discussed in substantially more detail than I have yet provided. Typical bull markets will cause the market to rise to a level that it has never reached before. Sometimes, these gains are so large that no bear markets ever take the market back down what had been that bull run. (For example, the market hit a high near 370 in 1990 and went on a bull run that lasted until 1998. The lowest value that the market has hit since then is about 680.) This is sort of an extreme case, but it illustrates a useful principal. It's really hard to predict how longer a market cycle will last, how high it will climb or what percentage of its gains it will lose... but it seems to be a pretty safe bet that the market will fall to less than double its previous peak again once it has started another bull run. The first pull back in 1998 never brought the market back down below 740. It didn't get close to 740 again until the internet bubble collapsed, and even then it only came down to 800. (If I was following this strategy, I probably would have interpreted that collapse as having satisfied this general principal, and bought back in at that collapse instead of waiting another 8 years for the market to actually fall below 740.)
So basically, the advised strategy is to treat anything as the formation of a bubble when the market goes above double previous highs without any intervening correction bringing them below, let's say 250% of previous highs, just to include a margin of safety.
This is a very bullish strategy. It practically never tells you to sell. For reference, the previous high before our current bull run is around 1500. This strategy says to stay 100% invested until the market crosses above 3775 (unless there is another intervening bear market that gives us a new cap to work with before the market hits 3775, in which case the cap that this strategy advises will be substantially higher still.) Actually, this approach is ridiculously bullish. Pretty much everyone expects the market to hit a correction before it reaches 3775, and they're probably right. However, if the market does happen to reach 3775 before the next correction, you can be practically guaranteed that pretty much everyone will be thinking that markets have changed and that this bull run is going to continue going on forever. That's exactly what happened in the dot com bubble. That's when you'd know that you want to start selling.
THIS IS NOT INVESTMENT ADVICE (simply a description of an algorithm): To turn these observations into a strategy, stay 100% invested by default. Keep track of the peaks of the all previous bull markets (where a bull market would be defined as any period of time in which there was no correction of 10% or more). Whenever a new bear market occurs, eliminate a point from your list of previous peaks if the low point of the market is less than 250% of that peak. If there is currently a bull market, and you still have at least one peak on list that has not been crossed off, and the bull market is greater than 250% of that peak, begin reducing your holdings. You want to do this slowly so that you will be continuing to sell as the market continues to climb. When a bear market brings the market back down to less than 250% of any peak cross the peak off, and buy back to full investment when all of the peaks are crossed off.
Chances are, this strategy will never tell you to sell again in your life time. There was a ridiculous bubble in the 1920's and the next one didn't start until the 1990's.
That's sort of the point.
The goal is to create a simple strategy that will outperform the market by identifying ludicrous bubbles. Only two such bubbles have occurred since good market data has become available, and this strategy would have identified both of them. (As for risks of overfitting, my gut-level estimate of what would be an unreasonably large bull run was for a run to bring the market to double its previous peak. The data did support this instinct, but it did provide as much of a margin of error as I would have hoped for and expected. So I've revised it to be a little bit wider.)
The only way that this stock market can underperform the market, is if at some point in the future, something equivalent to the stock market of today shooting past 3775 without ever having a correction of 10% between now (it's at roughly 2100, when the previous record high before this bull market was 1500) and when the market hits 3775. Obviously, today this sounds ludicrous... and if it ever stops sounding ludicrous, it's probably a pretty strong hint that there's a bubble which is distorting everybody's sense of how the market should behave.
NOTE: I am not suggesting that the markets are likely to hit 3775 before the next correction. I think that that is exceedingly unlikely. I expect the next major correction to occur before 2500. But what I am much more confident of is that the market should hit another correction before it climbs to 3775, and that if it doesn't, practically nobody will retain the appropriate level of pessimism that expects the market to com back down... if it does go on another insane rally like one analogous to the dot-com bubble, then it keeps climbing and climbing until it hits prices that no one today would call sane but that practically everyone would think are the new normal during the rally. To repeat something analogous to the dot-com bubble in a world where our most recent bull market topped at 1500, would involve the market eventually climbing to a level that tops at 7000 without an intervening correction of more than 10% that takes the market down below 3775. (That's the notable thing about the 1998 correction. When evaluated against what the market had been doing, it seems like a significant hair cut but when evaluated against the previous record high before the bull run started it looks like nothing happened at all.) If markets were to hit these kinds of ludicrous valuations, if they really do climb to 7000. Well then, everyone who started to think 3775 looked high will start to feel like they were missing something. It will be really hard to doubt the optimists. And if there is some little correction along the way pulling the market from 6000 down to 5000, say, then it will be really easy to believe that that was the correction, when it really wasn't. And this is one of the things that will be practically impossible to see at the time, even though, it is really obvious a few years earlier, and will become really obvious again five years later.
In this post, I will lay out the simplest strategy I can devise to outperform the S&P 500. Since picking stocks is anything but simple, I will not make this strategy a stock picking strategy. Instead it will be a market exit and re-entry strategy for investments kept at market weight in the S&P 500. Timing exit and re-entry is itself complicated, and not something that someone can do with a lot of confidence. So instead of looking for a way to completely exit and re-enter the market, I'll talk about reducing a position and strengthening it again.
If you're willing to go with an empirical strategy as opposed to one with more of a theoretical justification. Looking at the 100 period exponential moving average has actually been a remarkably accurate predictor of stock market collapse. Sell as soon as the market crosses below, buy when the market crosses back above. (But this strategy does have a reasonable theoretical justification. 100 periods because 100 is a nice round number -- which is evidence that I didn't tinker with the strategy in search of a perfect fit but just tested something "obvious" -- exponential moving average because averages are easy to calculate and returns are basically compound interest so using exponential as opposed to arithmetic averages makes sense. I have not actually looked at the arithmetic average to see how well in predicts anything, because it's arithmetic which means it really isn't mathematically suited to describe the market. I plan to use it, or at the very least, take it under advisement in my investing. UPDATE: I did tinker with the numbers a little bit, and you get better returns for periods in the range 30-60 than you do for a period of 100, but the smaller the number you choose, the more in and out there is, and the less significant the predictions are. Crossing in the 100 basically means the market is about to collapse, but crossing with smaller numbers doesn't indicate that nearly as strongly. If I was using a smaller number, I would sell as soon as it crosses below, but wait for it to remain above for 48 hours or so, and make sure it stays above for that whole time -- but the more detailed strategy has much more potential to be overfitting the data.)
Throughout this example, I'm going to be assuming that a person as a certain particular amount to invest, and does not have another source of income to use for adding money to the investment. I'll also be assuming that no one is deallocating money that had initially been set aside for investment. In the event that someone is additional money or removing existing money from the investment account, everything about this strategy should be re-normalized. (In other words, "selling" means reducing the percentage of money available to invest that is invested, and "buying" means increasing the percentage of money that is available to invest that is invested. If the strategy says to sell until 90% invested and keep the remaining cash on hand and someone adds another $10,000 to the account. $1000 of that should be added to the cash on hand, and $9,000 should be invested if the person is abiding by this strategy even if the strategy says not to by buying.)
The stock market is not efficient, nor is it a random walk. It exhibits certain consist patterns. In particularly, the market moves in bear-bull cycles. These cycles have a definite upward pattern. On average the highs of each bull market are higher than the highs of the previous bull market, and the lows of each bear market are also higher than the lows of the previous bear market. At the top of each bull market (or frequently even a year or two before it reaches the top of a bull market), the stock market hits record highs. However, during corrections, it does not hit record lows. In most corrections it doesn't even hit ten year lows. It has never hit a thirty year low.
From this observation alone, we have singled out a very specific market event that is worth noting, and can define a strategy that revolves around that particular event. The notable event is the market hitting new record highs.
Ideally, you want to buy stocks when the market bottoms out and sell them when it peaks, but timing this ideal is impossible. However something that is possible is to only sell when the market is at record highs, and only buy when the market is not at record highs. Of course, once the market has hit a record high, every new move upward is a new record high. Every move downward below any of those record highs are no longer record highs. Thus, it is still possible to underperform the market while only selling at record highs and only buying when the market is not at record highs. For example, the market could reach 2150. Someone could sell everything at 2160. The market could continue climbing all the way to 2500, before contracting to 2400 which would no longer be a record high. Someone could then re-enter at that point and underperform the market by 10% while abide by the two rules I've just given. So they don't define a winning strategy yet. Actually, since they don't give much definition about when to buy and when to sell, they aren't much of a strategy at all.
Actually, it's possible to be quite a bit more nuanced than this. If the market will ever hit a value below it's present value again, you benefit by selling so long as you buy back when it is at a lower level. The ideal time to buy stocks is whenever the market hits a level that is low enough that it will never reach that level again.
Bull markets and bear markets can be discussed in substantially more detail than I have yet provided. Typical bull markets will cause the market to rise to a level that it has never reached before. Sometimes, these gains are so large that no bear markets ever take the market back down what had been that bull run. (For example, the market hit a high near 370 in 1990 and went on a bull run that lasted until 1998. The lowest value that the market has hit since then is about 680.) This is sort of an extreme case, but it illustrates a useful principal. It's really hard to predict how longer a market cycle will last, how high it will climb or what percentage of its gains it will lose... but it seems to be a pretty safe bet that the market will fall to less than double its previous peak again once it has started another bull run. The first pull back in 1998 never brought the market back down below 740. It didn't get close to 740 again until the internet bubble collapsed, and even then it only came down to 800. (If I was following this strategy, I probably would have interpreted that collapse as having satisfied this general principal, and bought back in at that collapse instead of waiting another 8 years for the market to actually fall below 740.)
So basically, the advised strategy is to treat anything as the formation of a bubble when the market goes above double previous highs without any intervening correction bringing them below, let's say 250% of previous highs, just to include a margin of safety.
This is a very bullish strategy. It practically never tells you to sell. For reference, the previous high before our current bull run is around 1500. This strategy says to stay 100% invested until the market crosses above 3775 (unless there is another intervening bear market that gives us a new cap to work with before the market hits 3775, in which case the cap that this strategy advises will be substantially higher still.) Actually, this approach is ridiculously bullish. Pretty much everyone expects the market to hit a correction before it reaches 3775, and they're probably right. However, if the market does happen to reach 3775 before the next correction, you can be practically guaranteed that pretty much everyone will be thinking that markets have changed and that this bull run is going to continue going on forever. That's exactly what happened in the dot com bubble. That's when you'd know that you want to start selling.
THIS IS NOT INVESTMENT ADVICE (simply a description of an algorithm): To turn these observations into a strategy, stay 100% invested by default. Keep track of the peaks of the all previous bull markets (where a bull market would be defined as any period of time in which there was no correction of 10% or more). Whenever a new bear market occurs, eliminate a point from your list of previous peaks if the low point of the market is less than 250% of that peak. If there is currently a bull market, and you still have at least one peak on list that has not been crossed off, and the bull market is greater than 250% of that peak, begin reducing your holdings. You want to do this slowly so that you will be continuing to sell as the market continues to climb. When a bear market brings the market back down to less than 250% of any peak cross the peak off, and buy back to full investment when all of the peaks are crossed off.
Chances are, this strategy will never tell you to sell again in your life time. There was a ridiculous bubble in the 1920's and the next one didn't start until the 1990's.
That's sort of the point.
The goal is to create a simple strategy that will outperform the market by identifying ludicrous bubbles. Only two such bubbles have occurred since good market data has become available, and this strategy would have identified both of them. (As for risks of overfitting, my gut-level estimate of what would be an unreasonably large bull run was for a run to bring the market to double its previous peak. The data did support this instinct, but it did provide as much of a margin of error as I would have hoped for and expected. So I've revised it to be a little bit wider.)
The only way that this stock market can underperform the market, is if at some point in the future, something equivalent to the stock market of today shooting past 3775 without ever having a correction of 10% between now (it's at roughly 2100, when the previous record high before this bull market was 1500) and when the market hits 3775. Obviously, today this sounds ludicrous... and if it ever stops sounding ludicrous, it's probably a pretty strong hint that there's a bubble which is distorting everybody's sense of how the market should behave.
NOTE: I am not suggesting that the markets are likely to hit 3775 before the next correction. I think that that is exceedingly unlikely. I expect the next major correction to occur before 2500. But what I am much more confident of is that the market should hit another correction before it climbs to 3775, and that if it doesn't, practically nobody will retain the appropriate level of pessimism that expects the market to com back down... if it does go on another insane rally like one analogous to the dot-com bubble, then it keeps climbing and climbing until it hits prices that no one today would call sane but that practically everyone would think are the new normal during the rally. To repeat something analogous to the dot-com bubble in a world where our most recent bull market topped at 1500, would involve the market eventually climbing to a level that tops at 7000 without an intervening correction of more than 10% that takes the market down below 3775. (That's the notable thing about the 1998 correction. When evaluated against what the market had been doing, it seems like a significant hair cut but when evaluated against the previous record high before the bull run started it looks like nothing happened at all.) If markets were to hit these kinds of ludicrous valuations, if they really do climb to 7000. Well then, everyone who started to think 3775 looked high will start to feel like they were missing something. It will be really hard to doubt the optimists. And if there is some little correction along the way pulling the market from 6000 down to 5000, say, then it will be really easy to believe that that was the correction, when it really wasn't. And this is one of the things that will be practically impossible to see at the time, even though, it is really obvious a few years earlier, and will become really obvious again five years later.
Comments
Post a Comment