What has your hit and miss ratio been?

[damdin]

Interesting discussion guys, so keep it going (despite some contradictory suggestions (hold onto losers to turn it to winners vs close out losing position strictly etc))

 

[DeepState]

Trading shares long only without a trailing stop.

1 position - maximum loss is 100%
1 position - maximum profit is infinity

The same with a stop

1 position - maximum loss is < 100%
1 position - maximum profit is still infinity

Thanks again. This is good. I am learning from you, which is appreciated.

As you mentioned in an earlier post, what matters is expected returns. Quoting a min-max outcome boundary does not say much for the expected return. It just quotes the outcome boundary for a single gamble.

Let's make this really simple. Even simpler than before....which I didn't imagine was possible when I wrote it...surprised myself. Let's focus on a portfolio consisting of a single stock which moves with Geometric Brownian motion with drift. It can, as you point out, move asymptotically to zero, to drift on up towards infinity.

Strategy 1: Buy and hold a the stock...no stop-loss, no nothing
Strategy 2: Buy the stock with 10% trailing (or any kind of stop method for that matter, for any magnitude of trail or fixed or some algo) stop...replace the stock with another drawn from the stream of stocks in the market and repeat to infinite time horizon.

I suggest that, with respect, you aren't aware that the expected return from both of the above is actually identical (ignoring brokerage and trading frictions and tax realization which would actually favour Strategy 1 under conditions of a positive drift).

Once again, it obviously works for you. So I don't dismiss it because of divergence between belief and underlying empirical or theoretical observation. I should add that we are talking about a situation where leverage doesn't come into the picture.

Why else do you, or others, do this?

 

[DeepState]

Interesting discussion guys, so keep it going (despite some contradictory suggestions (hold onto losers to turn it to winners vs close out losing position strictly etc))

Those suggestions may seem contradictory, but they are not, actually. Although both are straw men. Both broad concepts, without taking it to the zealous extreme, are deep features of what happens in practice in professional management. But my question here relates to the pervasive use of these rules in retail. You see it all over the place in FX and share trading. I'm hoping this thread will allow us to explore this. Let's see where it goes. Thanks for starting the thread.

Might as well chip in something here about why I think it is useful. It is a practice which overcomes a finding made by Shefrin and Statman in 1985 about the behavior and false beliefs of investors under conditions of uncertainty. The finding was called "Disposition Theory" or the "Disposition Effect". Without training, we sell winners too early and hang onto losers thinking they will revert. It results in high HR. It's the same impulse that drove you to want a high HR. Thanks for volunteering to be a lab rat! This was built upon heavily by a guy called Terrance Odean. These findings, though, just highlight false beliefs that heaps of us have before being confronted with actual data. And even then, heaps will not be convinced by data or know how to use it and continue on to believe in investment fairies. That's why you can find an edge if you work at it - but it's not as if people are that stupid to make it easy for you.

MichaelD is into investment psych, from prior posts, so I'm glad we are in dialog. I am hoping others will jump in because a ton of people on this site are doing it.

 

[MichaelD]

Why else do you, or others, do this?

The stock market is in a permanent uptrend. Has been for 200 years.

That's why a portfolio of 20-odd stocks with a trailing stop is profitable. It won't work with 1 stock. You need 20-odd, appropriately position sized.

Limits the downside.
Doesn't limit the upside.

Individual stocks can go to zero.
A diversified portfolio won't.

A trailing stop can be almost anything, so long as the winners are allowed to run and the losers are limited in some way. It can be technical, it can be fundamental, it can be astrological. It can even be zero.

 

[DeepState]

The stock market is in a permanent uptrend. Has been for 200 years.

That's why a portfolio of 20-odd stocks with a trailing stop is profitable. It won't work with 1 stock. You need 20-odd, appropriately position sized.

Limits the downside.
Doesn't limit the upside.

Individual stocks can go to zero.
A diversified portfolio won't.

A trailing stop can be almost anything, so long as the winners are allowed to run and the losers are limited in some way. It can be technical, it can be fundamental, it can be astrological. It can even be zero.

Thanks for continuing with the dialog.

We have allowed for the 'permanent uptrend' in assigning a positive drift to the stream of returns.

In my prior post I highlighted that a single stock strategy which is left to float along in the positive drift is expected to perform equally well as one with stops in place using any of the strategies you have mentioned, even those with zero stop limit. These stops can be developed via any method including via random number generator or Fibonacci sequence. Hence, applying stop losses that let trends run whilst cutting losses on a single stock portfolio does not add value. You seem to allow for this in your latest response when you say it won't work for a single stock.

You go on to suggest that it would work for a portfolio of 20 such stocks and applying this strategy. It actually doesn't. If it doesn't work for one stock, building a portfolio of 20 or arbitrarily large number of stocks and weighting them in any scheme you like will not change the fact that stop losses of any description considered here will not generate value add vs the buy-hold scenario under the conditions we have described. It simply shifts the slug ratio more positively if the stops are tighter, but (as you have noted) damages the HR. The net result is no gain...ignoring t-cost. Since you've raised it, I'll go further and say that it doesn't even shift the overall risk vs a buy-hold portfolio as the expected distributions are identical...not just the return expectations. That is, the risk of effective bankruptcy are identical in both cases. Though ours is a toy world, any approximations to the real world on this issue due to jumps or gaps washes away with modest diversification.

There is no alpha generation in stop losses based on arguments related to
- skew
- outcome range
- portfolio diversification.

Yet...here it is...alive and kicking. It is clear that you believe there is alpha but the rationale provided (I am grateful, your reasoning matches with that provided elsewhere and is representative) actually doesn't bear out for the most part unless there are special conditions present which have not been raised. I am happy to stand corrected. I want to be corrected. Who wouldn't want to sit by a beach with stop loses on generating additional value over the market? Ship it in!

You are clearly experienced and it is very nice to correspond with someone whose actually been around a bit. If I may, where/how did you learn to apply stop losses? Why did you find the concept attractive from a psych viewpoint?

 

[MichaelD]

You go on to suggest that it would work for a portfolio of 20 such stocks and applying this strategy. It actually doesn't.

You are clearly experienced and it is very nice to correspond with someone whose actually been around a bit. If I may, where/how did you learn to apply stop losses? Why did you find the concept attractive from a psych viewpoint?

We'd have to agree to disagree on the value of a trailing stop. All the extensive backtesting I did way back in the past confirmed a dramatically improved result when a stop loss was applied.

Where did I adopt the practice of using a stop loss? I'd say I picked it up from several of the trading books I read early on and then verified what it did with a lot of backtesting.

I don't find the concept attractive psychologically. I hate a 40% win rate. I am, however, still in the game despite only trading long with copious leverage during the GFC. I also have a portfolio that traded exactly as suggested by backtesting - a string of small losses to begin with, and now almost everything I hold is hugely profitable.

In short - it has worked for me as I expected it to.

 

[DeepState]

We'd have to agree to disagree on the value of a trailing stop. All the extensive backtesting I did way back in the past confirmed a dramatically improved result when a stop loss was applied.

Where did I adopt the practice of using a stop loss? I'd say I picked it up from several of the trading books I read early on and then verified what it did with a lot of backtesting.

I don't find the concept attractive psychologically. I hate a 40% win rate. I am, however, still in the game despite only trading long with copious leverage during the GFC. I also have a portfolio that traded exactly as suggested by backtesting - a string of small losses to begin with, and now almost everything I hold is hugely profitable.

In short - it has worked for me as I expected it to.

Backtest! You are truly a goldmine MichaelD. With the ECB putting a ceiling on the EUR vs RoW your stock looks good indeed, particularly in EUR numeraire..

BTW, please recall that I use a version of stops too although the method wouldn't be recognizable. Believe it or not, there are elements in it that are distant cousins of trailing stops. So we don't disagree in philosophy at all or have to agree to disagree. Is that agreed?

When you conducted your backtest...what did you code for the action taken subsequent to hitting a stop?

 

[MichaelD]

When you conducted your backtest...what did you code for the action taken subsequent to hitting a stop?

Hit stop -> close position.
Open another.

 

[DeepState]

Hit stop -> close position.
Open another.

In the same stock? With equivalent size? With no delay?

 

[MichaelD]

In the same stock? With equivalent size? With no delay?

No, in a different stock.

Sell the downtrending one.
Buy an uptrending one.

Position size is 0.5% risk so depends on where the stop is and on overall capital.

 

[DeepState]

No, in a different stock.

Sell the downtrending one.
Buy an uptrending one.

Position size is 0.5% risk so depends on where the stop is and on overall capital.

Two fat ladies 88. Bingo. Welcome to the world of special conditions.

I can see where you are coming from and why your opinions and outcomes are as they are now. Keep it going. Thoughtful risk management too. But, ultimately, it's not actually the trailing stop loss that is creating the value other than, perhaps, acting as a form of alert. Despite the pain of a 40% HR, you clearly derive benefit from the belief that it protects you from going bust. That's totally understandable if you are investing with leverage. Leaving the issue of leverage aside, in reality, the trailing stop just helps you get what's available from your personal decision rule. ie. it works, but not for the reasons we have been discussing. The trailing stop is actually a rebalancing trigger for you and it happens to broadly suit the real thing that's making your money. On its own, it does nothing.

We can agree that you are making money. All power to you. You should consider going equitized long-short to make even more. I will have to agree to disagree with you if you are still of the view that the trailing stop actually MAKES money.

This is thoroughly fantastic to discover from a range of perspectives.

I appreciate this dialog very much. You've been most generous in your views and opinions. I hope we have many more opportunities to exchange. May the trend be your friend forever more.

Best.

 

[brty]

Watching this thread with interest.
2 opposing views.
A sell winners add to losers, ie re-balance.
B sell losers add winners (stocks in uptrend).

Both will work with the correct stocks.

Consider A, in a market decline where everything is going down (ie 2008), re-balancing gives more money to the biggest losers from the smallest losers.

Consider B all stocks get sold, stay in cash until market turns.

I prefer B, did B in 2008. From trading a 6 figure sum, lost less than interest for year.
I add to winners not losers, seems the opposite of most people.
Love Ed Seykota's rules.
1. Cut losses. 2. Ride winners. 3. Keep bets small. 4. Follow the rules without question. 5. Know when to break the rules.

Every stock is different.

Retired Young, you seem to be on a crusade to have everyone re-balance, why?

 

[burglar]

Interesting discussion guys, so keep it going (despite some contradictory suggestions (hold onto losers to turn it to winners vs close out losing position strictly etc))

Contradictory because people are different.
It makes an auction possible.
If everyone was the same, the price would stagnate.

Keep trying to work it out ... but don't go nuts! :

 

[DeepState]

Watching this thread with interest.
2 opposing views.
A sell winners add to losers, ie re-balance.
B sell losers add winners (stocks in uptrend).

Both will work with the correct stocks.

Consider A, in a market decline where everything is going down (ie 2008), re-balancing gives more money to the biggest losers from the smallest losers.

Consider B all stocks get sold, stay in cash until market turns.

I prefer B, did B in 2008. From trading a 6 figure sum, lost less than interest for year.
I add to winners not losers, seems the opposite of most people.
Love Ed Seykota's rules.
1. Cut losses. 2. Ride winners. 3. Keep bets small. 4. Follow the rules without question. 5. Know when to break the rules.

Every stock is different.

Retired Young, you seem to be on a crusade to have everyone re-balance, why?

Hi Brty and also Burglr

They are not actually opposing positions. But let me build it up.

There are extremely few pure arbitrage opportunities in the market. Nothing is for sure. So we move to statistical arbitrage. We try to find a way to predict stock returns that, on average, have a positive profit expectation over time.

1. If you found $10k sitting on the street gutter, would you pick it up? I sure would. I'm pretty sure 99% of people here would. There might be 1% who believe in Efficient Markets Hypothesis who'll say "if it were real, it would be taken already" or another few percent who have a bad back.

2. If you found a box which contains an unknown amount of cash, including the possibility of losing cash, but on average gives you $10k...to make this better, I'll say the max/min is $15k and -5k? I sure would.

3. What if you could be presented with many situations like 2. ? It will smooth your experience out over time and increase the chances of getting a positive result despite the real risk of individual draws costing you money. Would you take it then? Apparently, most people would not. How awesome.

Rebalancing is like 3. Yet strangely, from the dialog, no-one seems to want to take it. That $10k expecations box just swings on by. I come across as on a crusade because of the number of posts and in trying to align beliefs with the reality (at least as I perceive it to be...and also the entire professional finance community, but no matter). In reality I don't know any of you. So, I'm really not crusading. I have no dog in this fight. I am, however, fascinated by the beliefs being brought here and the arguments being raised. I have benefited from McLovin, Smurf1976, So_Cynical and MichaelD. We can disagree, but I am richer for the experience.

Let's move to rebalancing. Why does it work? Why does it generate a positive expectancy? Let's use a simple example drawn from the research paper Bouchey et al (2012):



You can buy Apple and Starbucks. By rebalancing between the two to a fixed weight shown along the horizontal axis your portfolio (the curved line) will always exceed the return from a buy and hold (the straight line) starting with the same weights. Depending on the initial weights, you can get huge gains relative to the unbalanced portfolio.

That's the Black Box in action....chug chug chug. Booooooring. Yawn.

When the universe was created and all the laws of physics were set, the laws relating to the profitability of rebalancing were one of them. Here is the equivalent of the E=mc-squared of rebalancing for your viewing pleasure (from same source).



This equation is water tight. Feel free to kick it around. We've had 30 years to do it and instead of being killed, there are entire institutes and huge firms devoted to it. I think the guy who started it all off was Robert Fernholz in 1982. If you want to investigate how to make money boringly via rebalancing, then you might want to check out his work on Stochastic Portfolio Theory.

So, have we achieved the point that rebalancing has positive expectancy and that it is not a small thing. Further, although it can be very sophisticated to implement if you want to move into the professional leagues, you can benefit quite readily by just robotically doing it very simply.

You can do quite well just rebalancing to some reasonable weighting scheme in all stock portfolios or in multi-asset class portfolios.

Everyone on this site is a RockStar and thinks they have some method that will slaughter these returns. They can't all do so, and expenses will be incurred in the effort by brokers, market makers, sec lenders and margin financers. They are laughing their pants off at the over-trading that goes on in the market. They have yachts. Most on this site will never get close to it. Yet the game continues. Such is the lure.

When you enter the market, the best starting point is that you are average. You have no idea who you are up against and your best guess should be they are as good as you until proven otherwise. This is Bayesian adjustment. What actually happens is we generally have overbuilt expectations, overtrade, cause excess volatility and feed the rebalancing process. In aggregate we will underperform the, say, equity market by a couple of percent after fees, expenses and tax. Roughly 80% will fall behind a simple rebalancing program.

But, ask a roomfull of people if they are above average drivers and 80-90% will say yes. Same thing applies here. Awesome. 20% will actually do very well. I could give you study after study, fact after fact, but a man convinced against his will is a man unconvinced.

Here is a stream of black boxes for you. If you want it, feel free to pick it up. You'll beat 80% of the market over time just by doing it. If not...that's fine.

Now, I get the sense that this idea is seen as an either/or. It doesn't have to be. Let's say that you are a stop the losses and run with profits kind of guy. Doesn't it make sense to look at the portfolio you are holding at any time and check that it makes sense and is not drifting into stupid concentrations? Rebalance the thing to reasonable target weights. It doesn't have to be opposing. The concept is simply to rebalance to some fixed weights...then let the magic just do its thing over time.

The same applies to any method to make money that anyone can concoct. It could be astrological, as MichaelD suggests, which will actually do as well as the average. Actually it might do better given planets and celestial objects move slowly from Earth's perspective and thus generate fewer trades.

I hope this responds to your query. Please feel free to continue the dialog.

 

[qldfrog]

I really like you facts about rebalancing.I had no idea rebalancing had a positive twist.
Thanks

 

[burglar]

Hi Brty and also Burglr ...

The contradictions are either real or apparent.
To newbies they are real, palpable.

To the connoisseur they are apparent and can be explained away with a white paper on rebalancing

Okay! I didn't fully understand it, nor would a newbie.

I come to auction with a buy order hoping the price will rise.
It is met by a sell order.
A fair assumption, it is from someone expecting the price will fall??

I would then hypothesize that there are investors and traders at work.
And further, I would believe that these are two different animals.

One is buying in belief that the company is undervalued.

The other is buying and selling with no regard to the value of the company.
He is trading on the expectation that he understands what the market is doing.

Not in disagreement, just continuing the dialogue!

 

[DeepState]

just continuing the dialogue!

Hi Burglar. Sorry about the typo on your name before. Someone flogged the 'a'. Payback is a bitch.

Thanks for pushing on.

There are buyers and sellers. They move prices RELATIVE to one another. It doesn't matter what their motivations are. They could be buying on expectations of a rise, trimming their portfolio despite continuing to anticipate a rise, market making, redeeming to fund a $30pa budget requirement.....

Let's focus on Apple and Starbucks. Let's say you have an 20% Flipper / 80% Shark strategy whose algo was written by the Grand Master himself. You can even add your own fundamental overlay and then check the star signs with Athena. You discover that you favour Apple over Starbucks and after running it through your Northfield optimizer using its multi-factor risk engine, figure that a 80% weight to Apple and 20% weight to Starbucks is the right thing for you.

Because you are a total RockStar, anointed by the Grand Master himself, it so turns out that Apple did outperform Starbucks like you said it would. Can I please hang out with you?

If you don't rebal, the return to the portfolio is just 0.8 x Apple's return + 0.2 x Starbuck's return. That can be read off the graph before, producing a portfolio ending value of $40. Yeah?

Now, here comes the magic. I couldn't believe it myself when I first saw this and had to double and triple take. Holy smoke, a money pump. Really? BS. Then I saw the maths and all was good. But let's try to explain what's going on behind the maths. Breathe.

Each stock can be assumed to just drift upwards, but wiggles along on its path. Apple drifts upwards faster. As the prices wiggle around, the portfolio weights shift away from your target 80/20 weights. Now, at the start, the wiggling for a month, let's say, can either tilt the portfolio towards or away from Apple's initial weight. In the next month, the wiggling can either bring the weights further away from the initial weights OR back towards the target weights. Got this? We are only looking at two months. There are four possible outcomes. Super simple. Up-Up, Up-Down, Down-Down, Down-Up.

If you don't rebal, you can never profit from the times when the wiggling brings the weights closer to the initial weights...when the portfolio reverts. Rebalancing prior to reversion will always make you money (prior to t-cost). You are buying low and selling high. Or selling high and buying low. Rebalancing or not rebalancing for the Up-Up and Down-Down scenarios produce the same expected returns. The expected return for Up-Up and Down-Down are the same in combination, so any weighting scheme you come up with produces identical outcomes in expectations. In this case the weighting schemes are whatever the drifted portfolio has become, and the rebalanced portfolio weights (the initial weights).

So, if you rebal, every time the portfolio reverts, you make money. Buy-hold never grabs this. If the portfolio does not revert, you are no worse off than buy-hold. Hence, as time passes and the portfolio wiggles around, coins just appear and fill your money jar.

Voila.

This process basically pumps profit from the wiggling. This wiggling is better known as volatility. Hence this process is commonly known as volatility pumping or volatility harvesting.

So...you can express your investment insight AND you can juice it a heck of a lot more than most would give credit for, by volatility harvesting. From the example provided your rebal portfolio grew to $60 and your Buy-hold la-de-da portfolio, although built with awesome insight, grew only to $40. Please note, you have to allow for t-cost.

The more stocks you have, the more extreme they wiggle about and the less they wiggle in harmony....the faster you buy a yacht for yourself if you don't already own one.

So, hopefully this has demonstrated how rebal generates coin. Also, hopefully, it is abundantly clear that there is no either/or which is necessary between applying rebal and having your own insights.

Hope that helps.



Disclaimer: In this and prior posts on this thread, I am not providing advice. I am not recommending the purchase and sale of any securities. I am offering illustrations of finance concepts in passing conversation. I do not know your financial circumstances. Please do your own work.

 

[burglar]

... There are buyers and sellers. They move prices RELATIVE to one another. It doesn't matter what their motivations are. They could be buying on expectations of a rise, trimming their portfolio despite continuing to anticipate a rise, market making, redeeming to fund a $30pa budget requirement ...

I was aware, but you say it so much better than me.
Since landing here, ASF, I have found out about some forces in the market which were obscure to me earlier.
Obvious to me now, from reading your posts, that there are even more forces.

... Hope that helps ...

I'm not in your league, but I would be interested in your view of averaging down.


I leave this post with a quote:

“The stock market by its very nature is designed for you to lose money. The rallies and reactions within any trend ensure this process is at work constantly. It is created automatically. The market behaves this way because it has to! The weak have to perish so that the strong can survive. Professional traders are fully aware of the weaknesses in traders under stress and will capitalise on this at every opportunity.”

Read more:
free download from “Master the Markets”, Tom Williams

http://www.tradeguider.com/mtm_251058.pdf

 

[Bill M]

If you don't rebal, you can never profit from the times when the wiggling brings the weights closer to the initial weights...when the portfolio reverts. Rebalancing prior to reversion will always make you money (prior to t-cost). You are buying low and selling high. Or selling high and buying low. Rebalancing or not rebalancing for the Up-Up and Down-Down scenarios produce the same expected returns. The expected return for Up-Up and Down-Down are the same in combination, so any weighting scheme you come up with produces identical outcomes in expectations. In this case the weighting schemes are whatever the drifted portfolio has become, and the rebalanced portfolio weights (the initial weights).

So, if you rebal, every time the portfolio reverts, you make money. Buy-hold never grabs this. If the portfolio does not revert, you are no worse off than buy-hold. Hence, as time passes and the portfolio wiggles around, coins just appear and fill your money jar.

Hey DeepState I am enjoying reading your posts. I am way out of depth trying to understand some of the mechansims of rebalancing so if you don't mind confirming something for me. How often do think rebalancing should take place? I think every 3 Months "feels right" but I am not sure. Secondly you say Rebalancing or not rebalancing for the Up-Up and Down-Down scenarios produce the same expected returns.

My question is, do you rebalance at your preset intervals (in my case 3 Months) or do you rebalance when you have major market moves? Which scenario do you think works best?

 

[brty]

Being out of the market at times is also a postion.

Taking an Apple vs Starbucks example is the same as taking any two stocks that always go up as an example, very poor in the real world.

Each stock can be assumed to just drift upwards, but wiggles along on its path.

NO!! You cannot assume stocks will always drift upwards. How does your re-balancing work on the Japanese market during the same time period of your apple and starbuck example??

If you start with incorrect assumptions, ie stocks always go up, you ignore the probability the assumption is wrong. The Japanese market is the exception that kills the assumption. Waiting more than 20 years for the market to return to normal is clearly to long.

I like, and have stated on numerous occasions that the Piotroski method is extremely useful is choosing from a basket of stocks, and yes that is re-balancing every year, but not with the same set of stocks, though it is a subset of all stocks.
Do I use that? Some of the time, especially when they go up in price, I'm happy to add.