What do people think is a realistic average return to earn going forward for an account of say under $2M for example. I know there are some impressive models out there, but is attaining 30% a year for 20 plus years realistic at this point. I mean 15% per year would have you at $2M in just over 20 years with $100k investment and no additional contributions. Just curious what people think with all the discussions of triple digit returns on models, what they realisitcally strive for, or if you care to share, what your returns have been.
30% may be do-able if you are starting with $100 - $200K … if you are starting with $2MM then that might be a little too ambitious. With starting capital of $2MM you’d have an ending value of approx $380MM if you compounded at 30% for 20 years. Not saying it’s impossible but that’s a tall order. I think as your capital exceeds $100MM you might have difficulty compounding at 30% for a sustained period of time.
I think 20-25%/yr might be do-able. Note: this is assuming that benchmark performance for the next 20 years is typical of most 20 year periods. Hard to know in advance.
Ed
Charles,
To answer your question, 30%/year over 20 years is probably impossible (with our tools), over the right 5 years, yes, like 2003>2008, & 2009>2014.
I was able to achieve over 100% for a few years after I joined P123 in 2004 using 3, 3 stock micro-cap Ports with very careful trades to avoid high slippage. By the third year I had grown my portfolio enough that even with 3 new, 5 stock Ports and the original 3, 3 stock ports (24 total micro caps) I was unable to find enough of the best microcaps to still gain anywhere near 100% (that was 2006). The next year (2007), after adding additional cash from other sources, I had to move some of the cash to small caps to find enough liquidity to avoid high slippage. And we know what happened that year and 2008. I was out of the market much of that time after a 27% overall loss in my P123 Ports from their high in November 2007.
After I reentered the market in March 2009 I achieved a little over 100% that year, but again I had to move up to some mid-caps (3, 5 stock Ports), small-caps (4, 5 stock Ports), and only 3 of the micro-cap ports. After additional gains in my portfolio, but at much lower returns, I Now have 3, 5 stock large-cap Ports, 3, 10 stock mid-cap Ports, 3, 5 stock small-cap Ports, and I still trade 2, 5 stock micro-cap Ports although they have a very low effect on my portfolio since they are only 4% of my funds invested in equities. If the 2 micro-caps gain 100% over the next year they will only add 4% to my portfolio. The only reason I still trade them is that they are the most fun and exciting to pick up the occasional 100% gainers.
So the problem with gaining 30%/year over 20 years is that after a while you will have a large enough portfolio you won’t be able to find enough high returning micro-caps or even small-caps to invest in without excessive slippage. So you won’t have any choice but to move up to larger and larger companies where the high returns are fewer and harder to find since they are where most of the big money is chasing. There just is not as much alpha in them to find. However, that is not a bad problem to have!
Denny 
With an account under $1-2 MM, 30% should be realistic. However, as you noted, 30% is not sustainable as a portfolio grows in size. My impression is that 15-20% may be achievable for a $10 MM portfolio.
For very large portfolios, you would start competing with professional asset managers, who typically target an equity market return plus 2-4% alpha.
This all assumes that the current alpha generation for quant models stabilizes somewhat, as alpha has eroded significantly since the early 2000s due to increased competition.
Thanks for the insights guys. It just seems that even with a smaller portfolio (<$200K) 30% gains are tough to come by without taking on a good amount of risk.
I do not think that 30% year is realistic for most people for the reasons below
Allocation: To maximize long term return you also must maximize your equity allocation and be comfortable with the drawdown that will eventually come. Very few people will have a 100% equity allocation and holding other asset classes over long periods of time will be a drag on returns
Taxes: 25-50% year in the US in a taxable account. This is a massive drag on returns. Most people have some taxable allocation.
Scalability: Denny previously covers this.
Skill: Skill either designing your own and/or selecting and sticking with a r2go strategy despite the probable temporary underperformance (look at the current r2go and you will see that some are outperforming and others are underperforming their benchmark) that will eventually occur and the ability to avoid data mining. Most people will have one or more underperforming portfolios at any point in time(assuming that one holds a basket of diversified portfolios) Underperforming portfolios will be a drag on returns.
Scott
Scott – I believe the original poster was interested in the performance potential of just P123 equity strategies, and not necessarily an entire portfolio that should include other asset classes like fixed income. And, taxes can be mitigated or eliminated if funds are held in tax-advantaged accounts.
Regarding skill, this is obviously a wild card, as different users have varying capabilities of not only building robust models, but also sticking with a good model when it underperforms (avoiding “selling low”).
Either way, I stand by my earlier comments that a before-tax annualized return of 30% should be attainable for relatively small accounts. Perhaps it may be helpful to run a poll - for example, what are the annualized returns people have realized over the past 1 or 3 years? And include a range of AUM?
I think another way of looking at potential average gain is to use the Book for the models under study and get the 10 year average from that. The question is, how accurate is the in-sample to OOS results for sim vs reality? Always the magic question. My experience is I have been running several R2Gs in combination with my own ports using real trades since last Christmas (I have been doing paper studies for over a year first with AAII SIPro and then with P123) and have gotten good results. My trades are close to my simulations. I am happy. I am careful about my trades and trade only higher liquidity stocks (ADT60>$1M and Mktcap>FMedian). All of my ports are averaging in simulation 25%+/year (since 1999) so over a 20 year period, I think I should be in the ballpark for at least 2x of SPY (say 15-20%+/year). Not a lofty goal but good enough for me and worth the time and effort.
Depends which 20 years. We’ve had a 30-plus year period in which, notwithstanding some recessions and other crises, the Fed has provided a massive tailwind that had the effect of pushing stocks forward at a vigorous pace. But considering where we are now in interest rates, the best we can hope for is no change, or most likely at some point, a meaningful headwind.
We’ve also had, mainly in the early 2000s, a golden age of quant platforms, when they were starting to become both powerful and accessible but before they became as widely used as they are today, so it’s now necessary to work smarter.
Also, the micro- and nano-cap parts of the market are becoming much less of a no-brainer than in the past generation. On the one hand, the “small cap effect” is for real, and its very well known. The problem, though, is a dramatic increase in the number of players chasing it. Institutional presence and analytic coverage of the tiny stocks remains less, but it is more than it was and is growing. So adding rank factors like Market Cap -smaller is better and close(0) - smaller is better, which can easily pump up a backtest, may not be quite as effective with real money as they once were. They’ll still more likely outperform than underperform, but in terms of numbers, it’ll be a heck of a lot easier to show prolonged 30%-plus returns in backtests and sims than to get it with real money.
The scaling effect has already been discussed.
Ultimately, there are plenty of market inefficiencies still out there, but the most powerful tool is a starting point grounded in reasonable expectations.
So the average 1 year return for R2G models with 1 year returns is 26.47% (median 22.07%). Seems to make 30% questionable.
But is this a reasonable way to answer this question? Is there significant survivorship bias? I honestly have not been keeping track. Maybe people are keeping the better models to themselves and one can expect better returns with their own models. How else can one look at the R2G data without the problem of selection bias? Probably there is much that makes this a bad way to look at this question: I would love to hear 100 reasons to think that I can expect 50% or more. And please don’t tell me it is an efficient market after all.
BTW the one year return for the S&P 500 is 19.93%. One ray of sunshine I am holding onto for now is that the one year return for the Russell 2000 is 11.33%. If this is the better benchmark for comparison, the returns for R2G models are respectable if not dramatic. Especially if you believe that the Russell 2000 will outpace the S&P 500 over long time periods.
The annualized return for the Russell 2000 over the Max period is 8.11%. If I can beat that by 15% as the average R2G port has done out-of-sample over the last year (i.e., 23% annualized total) that is not bad and possibly realistic.