DV2 Performance In Review
Below is a chart of the DV2 indicator, an oscillator that I designed in 2008, otherwise known as the DVB (DV Bounded) as popularized by MarketSci in July of 2009. As you can see performance is strong this year- up over 30% despite a deep drawdown in May during the flash crash. The DV2 is on par for a typical year which is consistent with the lacklustre volatility we have seen YTD. Once again it is important to emphasize that it can be very difficult to outperform simple but highly effective strategies. Most traders would be better served paying attention to the best 3-5 strategies to form a composite judgement than trying to make highly nuanced predictions based on 10+ inputs that may not be among the best available (or worse yet multi-collinear). That said, strategies come and go, and watching the equity curve performance is important as an indicator of future performance persistence going forward. Strategies that break down immediately out of sample and never regain form are likely to be overfit. So far the out of sample performance of the DV2 is very strong, which is not surprising considering it is fairly simple to construct.
CSSA 
hi..
this performance is from trading witch instruments ?
best
hi Daniel this is the performance trading the SPY ETF(S&P500).
best
david
I’ve been trading a strategy using the combination of DVO, DVI and DVDS since April, and I can confirm the results seen using DV2 alone. May was not a very nice initiation period, but the indicators have performed extremely well since the end of June. Buying the Oscillator and Mean-Reversion Indicators package for Amibroker was the best investment I’ve made in trading tools in my entire fifteen years of trading.
Hi Jack, thank you very much for the kind words and the endorsement. It is nice to hear that you stuck through May and reached full potential. Perhaps you could send me an email and share your story with me in more detail.
best
david
I’ll be happy to. I’ll send you a note to the e-mail address you’ve provided on your “about” page for the blog.
I can’t remember the rules here – go long when DVB > 50 and short when DVB < 50?
Hi Damian, hope all is well with you. The rules were long50.
best
david
Thanks David!
David
I’m confused by what “long50” means in the context of the DV2. I thought the rules for DV2 were to go long on closes below 50 and short on closes above 50. Is this correct?
I see essentially the same results using dividend-adjusted data from Yahoo (31% so far in 2010). I always run tests with an estimate of slippage and commissions (based on my experience with SPY). Including total costs of 0.12%/round trip lowers this this about 18% this year. Still great, but transactions costs are heavy for DV2 alone. Personally I am considering other filters to smooth the returns, and maybe using futures to lower fractional costs. At the very least this has been a great way to see the market.
Zack
To Al, yes that is the correct method. thank you best dv
To Zack, this is somewhat true- though .12% is much higher than our experience at the institutional level which is roughly .05% round trip for SPY. Similar rates can be achieved with Interactive Brokers and Tradestation. There are many ways to boost the returns per trade including deeper entries and exits and using other indicators.
best
dv
Thanks for the follow-up David, entries like this are paramount at getting the gears going within my own head. I must say, that both this blog and MarketSci have inspired me to do a bit of my own research. I’m currently looking into what I would eventually call a ‘integrated DV2’. Thus far, it has proved difficult for a variety of reasons. 1) I’m a recent econ grad, so among other things (minimal trading experience, unemployment, funding, time constraints, etc) my quant skills are lacking 2) I have no access to Matlab, so I’ve been teaching myself how to use a free program called GNU Octave to perform the necessary calculation and 3) having to buy minute to minute data has slowed me down a bit. Never the less, in the next 3-5 months I’m hoping to of accomplished something. The premis of the ‘integrated DV2’ is that the market tells us much more on a day to day basis than what the DV2 accounts for. I hope to be able to weight (using two metrics) the daily high and low (and previous) to accomplish a more accurate indicator, after integration of a daily chart. In my head it makes perfect sense, so hopefully I can build off of your intuition. But, as already mentioned, merely adding more layers of analysis doesn’t always deliver a higher return (but if I can improve on the risk adjusted rate, I’ll be a happy camper). Anyhow, I just wanted to quickly thank you for sharing your ideas with us, as you’ve managed to invoke some serious thinking by doing so in this blog. All that being said, given what is mentioned in this post (in regards to volatility), I was curious if you had researched using other instruments to trade using the DV2? Perhaps the a R2000 ETF, being a bit more volatile, could outperform trading the spy with the DV2? Or maybe a couple commodity ETFs even?
Chris, we have looked into this variant of DV2 and it is in fact more successful–though you need to take into account the time of day and volume of trades for intraday price bars to make it practical. This is due to the noise around certain periods and thin volume that produces less meaningful highs and lows. The IWM is performing better with the DV2 then the SPY this year, and perhaps this is the beginning of a new trend in the dominance of IWM vs SPY for swing trading. The EEM is still the long-term and short-term winner in absolute and risk-adjusted returns. Many thanks for your complements, it always makes me happy that I am inspiring new research avenues.
best
david
I think we there is a confusion over the rules and David needs to reclarify: Go long when DBV 50, right? When are the trades executed–at the close of the signal day or at the open of the next day? I have played around but I have not been able to replicate a YTD of ~30%…
kostas, the rules are to go long when DVB < 50. This is executed at the close of the same day you take the reading which in practice would be done at 3:58. The results are accurate I can assure you. hope all is well and good to hear from you.
best
david
David
What am I missing? I thought the bounded DV2 was similar to RSI2 in that it ranged between 1-100. So in this backtest you are going long only? And what does “DVB50” mean? Go long when the DVB is less than 50? Greater than 50? When are you closing the position? Sorry I just can’t seem to reproduce the results
Amibroker DVB results for 2010:
Buy = DVB 0.50
Short = DVB > 0.50
Cover = DVB < 0.50
trades made on the close.
YTD Return = 31.40%
CAR = 35.26%
MaxDD = 12.25%
MAR = 2.88
62.07% winning trades (54 total)
37.93% losers (33 total)
Correction to last post:
Buy = DVB 0.50
Short = DVB > 0.50
Cover = DVB < 0.50
One last try, and then I give up.
buy DVB 0.50
short DVB > 0.50
cover DVB < 0.50
I am thoroughly confused. How could
Buy= DVB 50 be correct? Are you saying to go long when the DVB is exactly 50? That can’t be correct
Let me try it one more time…
buy when DVB is less than 0.50
sell when DVB is greater than 0.50
sell short when DVB is greater than 0.50
cover when DVB is less than 0.50
Then, you would have gotten the performance numbers I posted, above.
Thanks for clearing that up Jack. So you are in the market 100% of the time?
Yes. I’m long or short 100% of the time.
By using DVO, instead of DVB, combined with DVI, DVDS, and a couple other things, the trade size varies, so I’m not always maximum long or maximum short. The trade size variability gives the overall performance quite a boost. I use data from the $SPY as input, but trade S&P 500 e-mini futures.
Right. I was referring to this particular backtest being either 100% long or 100% short at all times. Thanks for the clarification.
Hi David-
Have you been tracking the performance of the DV2 indicator recently? I’d be curious to know how it performed 2013 and so far in 2014.
Thanks,
Gordon