Can anybody present investable live book or port without market timing with st dev <15% and ret>15%?

P123 Designers and Users.

Can anybody present such a system for at least $5M daily trading capacity and 20+ holdings (or at least 10 holdings if nothing else, sector allocation <30%) ?
Out of sample performance for at least 1 year (live book or port).
If market timing is used then st dev should be lower. Turnover < 600%, the less is better.

Short stock or ETF system (even simulated with correlation to equity < -0.5) with at least $1M daily trading capacity and st dev <35% and at least zero ret (without market timing 1999-2015) is also highly valuable.

We can hope for further cooperation with the designers of such systems.

I suggest that you consider this 3 R2G Port book:

Stitts Wealth Management - S&P 500 / Short S&P 500; https://www.portfolio123.com/app/r2g/summary/1098263

Stitts Wealth Management - DJIA / Gold; https://www.portfolio123.com/app/r2g/summary/1101154

Stitts Wealth Management - SP400+SP600 / LT TBonds; https://www.portfolio123.com/app/r2g/summary/1101714

All 3 Ports have greater than 2 years of out of sample data. The book has 22% annual return, 14.5% Max DD, Sharpe 1.6%, SD 10.5%. The Book of these 3 Port is currently down less than 6% since its high in May.

An ETF Rotation strategy can meet these parameters. I’ve been trading a rotation strategy since 2006 and while I haven’t always achieved > 15% AR, the StDev is well below 15%. This port has been live almost 2 years:


Denny and Peter (and also Fred Piard), thanks for the fast response, appreciate it.

There are four sources of returns (risk/return decomposition):

  1. Beta - market exposure (cap weighted)

  2. Smart beta - factor indexes (value, quality, size, or momentum etc) in the middle of active/passive approach. New thing on the investment market.

  3. Exotic beta - catastrophe insurance, event-driven, arbitrage, short volatility, destressed etc. Compensation earned by investors for taking exposure to non-market risks or for providing liquidity. Very high Sharpe for stable market conditions and disastrous in other case.

  4. True alpha. Limited and very valuable. Idiosyncratic risk.
    A) Security selection. In our case stock picking.
    B) Beta timing or smart beta timing (market timing in other words).

So now I’m looking for point 4A in our P123 universe.

Denny. I saw these models. First two models. Very nice simulated performance (sharpe above 1, high liquidity, alpha etc). But out-of-sample alpha that I’m looking for is only 2%-6%. It doesn’t show consistency with sim performance.
Third model shows outstanding out-of-sample results but relies heavily on market timing rules (that’s not what I’m seeking in P123). Therefore generated alpha is not mainly due to stock picking strategy in this case. What do all these models show without any hedging (or lets say TLT constant hedging)? I think you can make very good forward working models assuming that it would be designed not for R2G purposes (I mean an obligation to show good simulated results for the past 16 years to the detriment of the future performance :).

Dwpeters. Never saw this model. Is it private live? Could you please provide additional info about your system? It looks very interesting. I need additonal info.

Yes, my model is a live private portfolio. Rather than share my portfolio I’ll point you to an excellent short paper on the general idea behind it, here: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1585517

It’s a powerful approach and rather than timing the market in the strategy, I just include some safe and uncorrelated asset classes in my portfolio. Although it’s struggling this year it was only 33% in stocks during the August drop and is completely out now.

yury - I prepared an unhedged book simulation here: https://www.portfolio123.com/port_summary.jsp?portid=1385660

I would like to address some of your points.

“It doesn’t show consistency with sim performance.”

If you have read any of my past posts you will know that I have never claimed that OOS should compare with backtest simulation. My backtests are “optimized”, OOS is of course not. I am perhaps the only person making this statement, and I have been severely criticized for it, although I estimate 99% of R2G providers optimize their backtests, some to extreme levels. It is not “wrong” to optimize one’s backtest, it is only wrong that P123 presents backtest as “performance”, as it is not. Is it a bad thing that OOS does not show the same performance as backtest? Absolutely not. What is important is that the development process, which includes optimized backtest, has extracted some level of alpha.

“But out-of-sample alpha that I’m looking for is only 2%-6%.”

My first question for you is how much experience do you have trading the markets? Clearly you have some knowledge but knowledge doesn’t translate to experience. Some of the statements you make seem to be naive. For example “and at least zero ret (without market timing 1999-2015)”. Without market timing I doubt you will ever get OOS with no negative return years. And some years you won’t beat the benchmark. This is life. To address your point regarding performance, ANY alpha for largecap systems is an achievement. This is not to say there are not issues with my models, and will be addressed over time. I am mostly waiting for the new R2G to come out, it seems to be taking an eternity.

“There are four sources of returns (risk/return decomposition)”

So your assessment of sources of return is very impressive. However, this pertains to the individual port (model) level. When considering Books, it is a completely one-dimensional argument. The fifth source of alpha is diversification from multiple independent strategies. Use of differing hedges with differing market timing, if done correctly, is an advantage over one model or three models with no market timing.

"I think you can make very good forward working models assuming that it would be designed not for R2G purposes (I mean an obligation to show good simulated results for the past 16 years to the detriment of the future performance "

I have been arguing this for a long, long time. However, don’t make the assumption that the market timing for three models is strictly there to impress people. As you pointed out, the OOS results are better with the market timing than without. And this is with Gold and TLT, neither of which makes a great investment vehicle at this time. I use these because P123 doesn’t provide other ETFs as alternatives (such as currencies).

Take care
Steve

Yury,

You must not have created a book simulation of the 3 Ports I suggested or you would not have made that statement. of the 3 Ports, the last one to be launched was on 09/11/2013, A little over 2 years ago. since that time the book of the 3 Ports is up 24% (that’s an annual return of 12%) while the S&P 500 is up only 11%. The book has a max DD since launch of 7.5% and the S&P 500’s is over 12%. So even with the lousy market we have had over the last 2 years That looks like pretty good alpha and low drawdown to me.

The liquidity of the 3 Ports is $11.9M, $19M, & $92M (you asked for $5M), and the turnovers are 248%, 171%, & 656% (you asked for < 600%), and the 3 Ports hold 57 stocks when fully invested (you asked for 20+). So what do you really want?

Hi Yury

This R2G should meet all/most of your criteria.
https://www.portfolio123.com/app/r2g/summary/1224467

Cheers
whotookmynickname

I’ll throw three systems into the mix:

  1. (Attachment). This is my broker statement for an account that trades 2 ETF systems (6 ETF’s in total) with huge capacity. One system is mean reversion and one is momentum. This is live trading on my money YTD (I’ve only been trading these since year start). They meet your return/risk objectives. They are similar to the ETF systems I offer in R2G, but have somewhat higher turnover (300% and 700%) - and slightly more ETF holdings.
  2. See also:
    https://www.portfolio123.com/app/r2g/summary/1245612
    This is 80% hedged largecap value-quality system based on stock selection. It meets your objectives.
  3. I can also offer 2 systems on Quantopian that are market timing in both SPY and US sectors. They can be auto-traded. Quantopian is better for these types of systems as I can a) create books of them and b) vary leverage and exposure rates and c) import outside indicators.

I would also agree with Steve’s comments about a basket of systems being better (and also with his observation that long-only stock selection systems are always going to have DD’s in down markets, and that P123 is likely not the best place for finding long-short or shorting systems).

EDIT: Steve has also put in a ton of time building his systems and sharing his thinking and development process. I believe he can give you hedge fund quality long-side stock selection at a fraction of the cost. That may or may not beat the market in any market cycle - but no single manager is a lock to do that.

Good luck.
Best,
Tom


BROKERACCOUNT1.pdf (131 KB)

Yury,

Steve’s 3 R2G Ports I suggested use different market timing approaches and different hedge instruments. The reason I suggested these specific Ports (I don’t sub to them) is that they follow many of the same ideas that I am following with my own Ports. I currently am trading 11 five and ten stock Ports. 8 of those use various timing approaches. They started moving out of the market in March, some added hedges in June, July, and August, and now all but 2 are either in cash or hold various ETF hedges.

From Steve’s descriptions:

Stitts Wealth Management - DJIA / Gold; “switches between 7 DJIA stocks (approx.) and the GLD ETF depending on season and market characteristics.”

Stitts Wealth Management - S&P 500 / Short S&P 500; “holds 30 stocks (approx.) from the S&P 500. When the 20 day delta of estimated S&P 500 earnings go negative then a hedge is put in place consisting of short S&P 500 (ETF Symbol: SH) with a ratio of 40% of total equity.”

Stitts Wealth Management - SP400+SP600 / LT TBonds; “holds 20 stocks (approx.) from the combined S&P MidCap 400 and S&P SmallCap 600. Depending on season and direction of T-Note yield, the model will switch entirely to long term TBonds (TLT).”

You can see from the charts below how they move into and out of the hedges, frequently at different times.



Hello, everyone.

First, Stittsville123, take a look on your book risk measurements tab. P123 doesn’t show alpha correctly.
For example, your first unhedged port S&P400/600 shows 23 ret 0.83 beta and 2.25 alpha for the trailing 3 years. It can’t be so.
23-0.83*9.31ret for SPY = 15% alpha (assuming zero risk free rate). Refresh book models risk tabs on monthly basis and it will be recalculated properly.

You can achieve alpha through diversification theoretically even combining zero alpha ports. It depends on each model st dev, correlation to the market, pair correlation and model returns. If combined beta decreasing faster than returns you achieve alpha and vice versa.
In other case you can decrease your alpha through diversification even using positive alpha models. It should be calculated case by case.
Then you using market timing rules correctly your beta becomes your alpha.

Bonds or shorts have inherent alpha due to negative beta. That’s why using it in a book brings diversification benefits. Using low correlated models in a book gives the same result. But the problem is that long equity will be positive correlated anyway.

Short tradable system with zero return over 15 years (it means that alpha is a positive number and equal beta*market ret) is possible to make even without market timing. I have attached one simulation and it works as live port now.

Backtested and OOS perfromance should be consistent. In other case it would be good-looking pictures only in R2G section without any practical use. Especially with 5-10 holdings ports. How many of these ports showed consistent OOS taking into account survivorship bias 10% or 5%? Pessimist addition :slight_smile: - it can be pure luck, it even should be luck)
The motivation to show good backtested result makes future performance significantly worse.

About my experience and what I want. I’m not a professional trader. My primary focus is private equity deals. But now there is an idea to launch a small hedge fund. P123 is the best place (taking into account price/quality trade-off) to simulate some ideas.



Denny, on presented 3 ports book without market timing (Stittsville123 post) OOS result (2 last years) looks almost the same as market and is not consistent with backtested result.
“The liquidity of the 3 Ports is $11.9M, $19M, & $92M (you asked for $5M)” I need at least $5M of my system daily trading capacity. It means 5%, maximum 10% of daily volume which should be at least $50M then.

whotookmynickname
You describe interesting approach, but it shows high cyclicity. However it uses market timing.
What is a clean result.

Yury - I don’t have much to say positive or negative regarding most of your post as I’m not really sure what points you are trying to make.

“Backtested and OOS perfromance should be consistent.”

Best of luck with that.

Steve

So… I am wasting time on here again… but

Many if not most hedge funds ‘market time’ and vary exposure based on perceived market cycle location and ‘risk.’ I read their newsletters regularly. It’s what most of them do. A basket of systems doing it systematically with rules is likely to be at least as good as ‘great pros’ doing it on a discretionary basis.

Here’s the hedge fund index for a dose of sanity:
http://www.barclayhedge.com/research/indices/ghs/Hedge_Fund_Index.html

0% returns this year, 3% in 2014 and 11% in 2013. They are doing market timing, hedging, long-shorts, varying exposure, etc. They have gotten crushed by a 60/40 portfolio of passive ETF’s over past 10 years. If you want hedging and all that, and you think you can regularly be in the top 10% of all hedge funds after 1 year or two of model building, good luck.

Here’s the CTA index:
http://www.barclayhedge.com/research/indices/cta/sub/cta.html
Down about 1.5% in 2015. Or the BTOP 50.
http://www.barclayhedge.com/research/indices/btop/index.html
Also down about 1% this year. Very mediocre since 2009.

Here’s the market neutral index:
http://www.barclayhedge.com/research/indices/ghs/Equity_Market_Neutral_Index.html

They have sucked for the past 10 years. Not bad this year at + 3.5% or so. But would have been better off in short-term bonds (treasury and corporate) over past 5 and 10 years with less risk.

I also did attach one of my broker account statements for real results. The 2 ETF trading systems I am using are up 2% or so this year. That’s about 3% better then the top 50 CTA’s (the BTOP 50) or 3.5% better then the CTA traders index. Doing the same thing with P123. And I still have control of my money. I think that’s a pretty good year considering. A 50/50 portfolio is down about 2.7% YTD (AGG/SPY). I am targeting 40% equity, 70% bond risk. On a total portfolio level, I’m down about 1.4% YTD. Not a great year. Not horrible. So, I am not sure why so many people continue to say that P123 is no good for ETF rotation systems or market timing? It’s at least as good as most of the professional alternatives.

Tom, your real trading result is great especially assuming current market drawdown and it is the final argument in all arguments:)
Hedge fund indexes is a tricky thing. You can find there opposite evidences depending on your point. Non investable indexes show not bad results.
There is consistency in generating alpha over time if looking properly in 10%-15% of funds.

Besides we have competitive advantage of a small size and low research and execution costs (read - team salary).

Market timing is possible, I agree, but it is very difficult to do properly (I mean it should work forward not for backtesting picture purposes) using only P123 tools. That’s why I’m seeking clean performance, 1) clean alpha without beta timing and 2) as you noticed diversification alpha that means negatively correlated assets which finally means shorting.

Also if you use market timing in long/short SPY port only it will give you two or more times increase in sharpe.
But if you use in a zero beta book (pure alpha generating book) it will give only 15-30% increase taking into account new transaction costs from dynamic allocation.

Can you show me strong, well conducted studies showing persistence in top managers performance over time periods of more then 1 year? Most of the studies I have seen show very short term outperformance (over following 1-3 months) by top hedge fund managers, declining significantly over long-time horizons (1-3 years). But some real value in minimizing big DD events during severe economic shocks.

See, for example:

  1. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=26509191. The most exhaustive paper I have seen - a meta study.
    Meta study on hedge funds covering 100+ papers published over the past 10 years. See pages 20-25 in particular. In general, the studies are very mixed but slight evidence showing that there is some very short term performance persistence, but it is weak. For some younger fund managers and for some of the studies there is persistence at a 3 and 5 year time horizon (but these studies are in the minority). Many studies do show that smaller stocks held by the very best hedge funds do outperform -particularly when managers are making concentrated bets. There is also some modest market timing and factor timing ability among funds that shows persistence. In particular, the best funds tend to reliably manage large ‘downside’ risks better (avoid huge DD events during extreme liquidity crisis). But, their market timing is very weak in other events. (My note: Their ‘left tail’ trimming results can likely be replicated at much lower cost, with more reliability through a basket of ‘automatic timing’ systems - such as many people here build).
  2. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=190389 (some persistence out to 3 months, but fades after that. They also show that most of the performance persistence relates to the lowest decile continuing to do badly). No evidence of ‘market timing bets’ aiding in persistence.
  3. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=489441. 1 and 2 month persistence. No evidence of longer term persistence.
  4. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=891169 . One of many studies presenting evidence that hedge fund managers ‘cook the books’ and ‘smooth out returns’ including inflating December returns to boost their bonuses (holding gains until later in the year).

Here is one of the few showing strong performance:

  1. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=686604. Using a relative ‘alpha t-statistic’ on trailing 3 year period to rank top managers produces performance persistence in following 3 year period. Specifically, about 25% of the outperformance from trailing 3 year period will spill over into following 3 year period (one of the major factors is that best managers attract more money, and more money lowers results). See pages 41-43 for the relative sizes of the alpha (.4 range). Even for the ‘top 10%’ the alpha, while statistically significant is very, very small relative to peers.

Any private funds that you are aware of that I should look at - funds you think are small still with earlier stage managers and lower AUM’s and are in the top 10% for their relative styles?

Agree that in theory, you could find long-only stock selection skill on P123 - but would likely want systems with large number of holdings (as most of the systems likely have large number of factors).

My proposal is very simple and based on common sense:

  1. OOS should be more or less consistent with simulation. Otherwise designers continue to waste their time on unproductive curve fitting with very small practical use.
    R2G models should start from at least 10 tickers (20 is much better). 5 stocks model OOS performance is completely unpredictable and dangerous. Good OOS result is due to luck only assuming huge survivorship bias.

  2. Separate alpha generation from stock picking strategy and market timing. (for example, present hedged and unhedged result)

  3. Present a rolling test in R2G.

  4. Allow short systems and whole books to be in R2G universe.

Yury -

Lets start with your “proposal” then we can examine the “common sense”.

What is your proposal? You have listed a set of requirements that are apparently important to you. Are you going to subscribe to an R2G that fits your requirements? Or buy a system from someone that fits your requirements? What exactly is the real purpose of this thread? If I didn’t know better, I would think you are planning an R2G launch and doing some advance marketing. (My spidey sense is tingling.)

Lets address this right away. Hedge fund managers give ZERO credit to simulations. If you want to be in the game then you need several years of audited track record. THEN you can line up with the thousands of other wannabe hedge fund managers.

Simulation is a development tool. How it is used is up to the individual designer. I expect that in the future, simulation will disappear from the R2G display, at least from sort and filter applications. The ultimate goal is positive buoyancy, not matching a simulation. The markets are constantly changing.

R2G subscribers are like my old girlfriends. They chase whatever moves the fastest and quick to abandon what isn’t providing continuous performance. R2G designers have pretty well demonstrated that they can put up impressive sims as fast as the previous models fail, so yeah, lets get rid of simulation completely.

10 or 20 (, 30, 50) stock models all suffer from the fact that all the stocks are based on the same criteria (buy/sell rules, ranking system). It has been my experience that if a 5 stock system performs poorly, then a 50 stock system will also.

You are again exhibiting “one dimensional thought”. A book of 10 systems with different buy/sell rules, different ranking systems, and different stock universes will almost invariably outperform one model of 5, 10. 20 ,3 0 50, any number of stocks over the long haul.

I’m not sure how you design systems, but the best way is to design the market timing to be compatible with the long stock portfolio. To do so independently is like implementing independent buy and sell rules. It can be done but doesn’t make a great amount of sense.

The rolling test is a good (well OK) test, but should be restricted to development use. It should not be a marketing tool.

So your liquidity requirements are for $5M+ dollar-volume yet the short simulation equity curve you showed was for a minimum 200k “clean 4000” universe. First of all, this short system doesn’t meet your basic liquidity requirement. How long have been actually trading this system? It is not good enough to watch OOS for short systems, you need to trade it to find out the limitations. What borrowing cost are you incurring? What slippage? Are you actually able to short the stock picks? What will happen in a bear market when the regulator outlaws short selling? - as in China today, and the USA back in 2008 (financial stocks).

Take care
Steve

“If I didn’t know better, I would think you are planning an R2G launch and doing some advance marketing. (My spidey sense is tingling.)”

Doing some advance marketing, that’s really funny. May be:). In fact I have already launched several R2G. I don’t anticipate any subscribers cause it consists many tickers. But for the independent track record that would be not bad.

The purpose.
I would subscribe to a system (diversified book) if it satisfy all my requirements. At now I’m comparing my system to existing in P123.
In addition it can bring some serious money into P123 from greater number of users and other user types (It means investment grade systems should be available in R2G)

The difference between 5 stock system and 50 stock system is in st dev primarily. And it is main concern. If you want attract real money you should minimize it.
Second, 5 stock OOS performance is completely unpredictable and even worse because of extreme backtest optimization already mentioned.

About books. I confirm that alpha can be achieved through diversification even combining zero alpha models in some specific circumstances.

About alpha decomposition on stock picking and beta timing. It should be done somehow (that’s the task for P123 team, they can find a way how to do it properly) because almost all more or less credible and investable models presented in R2G achieve primary part of their alpha from market timing (which can be done very simply in simulation).

Rolling test in R2G. Why don’t you want it to be in R2G? :slight_smile: We’ll see the real worth of 5 stock models? :slight_smile:

My short system.

Universe starts from $200K but in the buy rules I start from 300K. My $5M daily capacity assuming 5% or 10% VWAP applied for a total book not one model. It means I can move positions for $5M daily. Short stock model comprises only 5-10% of the total book. It means it should have at least $500K daily. 50 stocks in the model x 300k min liquidity x 0.05 VWAP = $750k

First gen of my short port I traded by myself. Live performance starting july I attached to the post.
Borrowing costs in the models - 7%. Variable slippage, hi and low avg prices. 10-15% of stocks are not available at the moment but becomes after several days. IB sends emails.
I checked these not available tickers, they bring the same avg trading result as the rest available tickers after fees.

Regards :slight_smile: