Announcing Quant Models Volatility: C2’s most profitable strategy (according to C2's hypothetical results)

Trading and Markets
1

Yesterday I made public a new volatility strategy: Quant Models Volatility.

Using Collective2’s hypothetical performance results, based on total return Quant Models Volatility is the most profitable strategy currently on Collective2.

First, the track record on C2

My strategy is 82 days old.

Among all strategies more than 60 days old, Quant Models Volatility has:

  • The highest total return (2213%),
  • The highest annualized return (with or without trading costs),
  • The highest 60 day return,
  • The highest 30 day return,
  • The highest Sharpe ratio,
  • The highest Sortino ratio, and
  • The highest Calmar ratio.

Among the 10 strategies with the highest returns on the Leader Board yesterday, May 9, Quant Models Volatility has:

  • (again) The highest total, annualized, 30-day, & 60-day returns,
  • The lowest maximum drawdown (13.0%),
  • Tied for the lowest Starting Unit Size ($5,000), though I would recommend a minimum of $10,000,
  • The lowest monthly subscription fee ($77).

Indeed, among the 20 strategies on yesterday’s Leader Board with the highest returns, Quant Models Volatility has the lowest subscription fee ($77).

++++++++++++++++++++++
Launching the strategy after 82 days

Even though I have had the highest total returns on C2 for nearly a month now, I did not want to take advantage of that simply to get early subscribers. If this was a good strategy last month, it should still be a good strategy this month—or next month. I wanted to show evidence that my strategy works for at least another few weeks. Indeed, I was so reluctant to try to enroll subscribers before 90 days had passed that I had posted on this forum that I intended to wait 90 days before going public. But with a good period for volatility trading in progress and a signal day possibly coming up very soon (perhaps just before the 90th day), I thought I should go public about a week early.

++++++++++++++++++++++
Notes on my Largest Drawdown so far

According to C2, my largest hypothetical Drawdown (13%) occurred on March 8 and 9.

  • My account closed on March 7 at $2236.
  • It went down 3.4% on March 8 to $2159.
  • On March 9, it went up 1.9% to $2199.
  • On March 10, it went up another 6% to end at a new closing high of $2331.

So the end-of-day drawdown was only 3.4%, but C2 uses intraday balances, so the C2 hypothetical DD is substantially higher than that (13%).

++++++++++++++++++++++
The Basic Timing Model & Other Trades

I have backtested my basic timing model for the period from April 2004 to March 2017 (100% XIV v. 100% cash)–with no adjustments for costs. The extremely hypothetical annual compounded returns for the model were 83.1%. The max drawdown for this hypothetical model since 2004 was 32.5%. Given that this model was optimally fit to past data, it is more useful for developing trading signals than for estimating actual future returns and drawdowns. Further, these hypothetical returns are unrealistically high because they do not include slippage, trading commissions, subscription fees, or autotrading costs.

I developed the basic timing model from backtesting in September 2016. Since then, it has performed much better than in the period from which the data were generated, which is rare. But then, this has been an extraordinarily good period for volatility models.

This timing model is based on several different kinds of indicators, but measures of contango are the most important.

To supplement my basic timing model that buys XIV, I do a number of things. Most common going forward will be to buy options, mostly to hedge a long XIV position, but occasionally to increase a long XIV position.

While my backtesting is unhedged, the concerns over drawdowns at C2 have persuaded me that in some situations I should use hedging to reduce the pain of black swan events. The hedging will take the form of investing less than 100% of my account in XIV, or buying out of the money UVXY or VXX calls, or buying UVXY. All these hedging approaches can be seen in the last 2 weeks of trading. I will, however, occasionally be long XIV more than 100% without any hedging.

Remember, trading is risky, and investors can, and do, lose money.

1 Like
2

Looks very nice return n acceptable risk. Right now, we are in different environment, we have the lowest Vix in 25 years n extremely low volatility. Most likely will reduce profitability in the future because there is high risk in front of us, only time can tell. If you use full capacity, will back fire when there is a big spike, if don’t , will decrease ROI.

3

Looks impressive.

If I understand you correctly, all the period from Apr. 2004 to Mar. 2017 was used for backtesting, and the out of sample results are only the 82 days that are displayed here?

1 Like
4

I wrote:

The basic timing model was developed in September 2016 and I have been trading it in my personal accounts since then. The out-of-sample performance I was referring to was in my personal accounts since September 2016, or just under 8 months.

The results shown for Quant Models Volatility since I started on C2 on Feb 16 result both from this basic timing model for XIV and from other discretionary trades.

5

I see. Another question - you mentioned discretionary trades. To what extent do they contribute to the system you run here?

6

I understand the advantage of a completely non-discretionary model, but I specifically used the word “discretionary” to disclose to you (and to David Stephens, who I think has raised this concern) that my strategy involved discretion. For various reasons (including trying to keep the DDs moderate and to partially implement some secondary indicators), my strategy is party discretionary.

There are two senses in which my model is non-discretionary:

(1) The basic XIV timing model is automatic in that it predicts when to be long XIV. But how I use the model is somewhat discretionary, particularly on the addition of hedging, which I have been adding to try to keep drawdowns reasonable. So it’s part automatic, the direction of the trade and whether to be in the basic position, but discretionary on how large the position should be and its precise makeup.

(2) I do have a firm percentage single day CLOSING drop in XIV prices that causes me to entirely close out a long XIV position. I don’t want to release what that is. I do not usually use traditional trailing % drops from recent highs.

Going forward, the main discretionary decisions are how to hedge and how large a long position in XIV I should embrace during periods that the automatic model says to expect rising XIV prices. How large a long XIV position to adopt is mostly based on a confluence of statistical indicators and their strength, but still involves discretion.

During the period my strategy was private, I was extremely successful in writing UVXY puts that were very temporarily mispriced. It remains to be seen how much I will be able to scale this part of the strategy up with subscribers because of the way that C2 handles limit orders (which once filled for anyone become market orders for everyone else). Every one of my trades selling UVXY puts traded through my limit price, usually far through. Still, I stopped writing these on April 25th, when I realized how limit orders in autotrading actually became market orders for many (or most) subscribers on C2.

So, if you exclude the period with the extraordinary (discretionary) returns from selling UVXY puts, the returns from Quant Models Volatility at C2 are still extremely positive.

If you consider only the two periods since Feb 16 during which I wrote no UVXY puts at C2 (Feb. 16-Mar. 20 & Apr. 26-May 9), my hypothetical daily return was 0.85%, which projects to an 8-fold increase over 252 trading days in a year.

Obviously, such a high 8-fold annual return would far exceed the results of my backtesting, and thus should not be expected going forward. But these two periods reflect trading mainly the main timing model, with discretion used to determine the size and precise makeup of the directional position and any hedging to be used.

7

You seem rational but how do you expect any rational person to believe you grew your strategy by nearly 1000% in a single month by investing in a reasonable manner? Assuming you can prudently invest in a way that you would earn 1000% in a single month without taking undue risk why on earth are you even bothering to market your system instead of trading it privately? Even 1k invested would suffice to become a billionaire within a year or two.

3 Likes
8

Interesting explanation on the puts. How were they so grossly mispriced that you could extract such profound returns from them so quickly? Looks like you used them to do a synthetic long with massive position sizing, but what if they went the other way and you quickly busted your account?

And why isn’t this strategy TOS?

Honestly this just looks like a clone of another popular vol strategy that used C2 weaknesses in handling limit orders on orders to create insane returns that can’t be replicated if any subscriber actually followed that trade (as you even note). While the strategy may be sound, I’m suspicious of such marketing tactics and the need for creating outsized (but completely unrealistic) returns that can’t be replicated just to lure subscribers.

2 Likes
9

I smell stalled quotes , more later .

10

From what I can tell, the open interest on the UVXY options is usually no more than a few hundred contracts per strike price. How is it possible then to write options in the tens of thousands of contracts on a single strike price if that far exceeds the number of contracts available? Am I missing something?

1 Like
11

First, I never claimed that I could do 1000% a month going forward. That would be ridiculous.

But that is not as good as my best month’s return that I’ve had in real life. I once had an 18-fold return in an option account in one month, my account going from $5,000 to $90,000 in that time. Needless to say, I never expected that sort of return to continue–and it certainly didn’t.

I would be ecstatic and pleasantly surprised if I could meet my long-term goal:

**

  • In the period after I went public, having a higher long-term return than any volatility system with a lower darwdown and having a lower drawdown than any system having a higher long-term return.

**

Let’s separate two sources of the high returns in this strategy: the basic market timing model and the selling of naked temporarily mispriced UVXY puts.

I think I have an excellent market timing model. In my own IRA accounts, I have been using this model since September 2016, and I have indeed almost certainly made more money already trading that model for 8 months than I could ever make on this site. And I have a large IRA portfolio currently devoted to buying XIV and volatility options.

As an academic, I have always considered developing market-beating strategies to be one of the most difficult intellectual challenges in the world. And I make a good salary as a professor, so the motive is partly intellectual, as well as partly financial.

As to writing naked UVXY puts, when in March I noticed the pattern of trading that I wanted to exploit, I had only pension and IRA accounts that did not allow writing naked options, so I couldn’t have easily implemented the strategy even if I had wanted to. (Buying 35% of a house for my daughter last year cleaned out my cash accounts.) Also, when I started writing naked puts, I wasn’t certain the strategy would work.

If In mid-March I had had access to a cash account allowing selling naked puts instead of my owning or managing sizeable pension and IRA accounts without such privileges, I might well have implemented this strategy for myself then.

Without this access, I decided to try it in my C2 strategy account. And it worked. Every single trade that was filled “traded through” the limit price I had placed, which is a higher standard than the usual one here at C2 that allows trades at the limit.

C2’s announced policy is best laid out in this post by Matthew Klein:

,

12

My first sale of UVXY puts was on March 21. After rescaling, I sold short 2.6 put contracts. 60 contracts at that strike at that expiration traded that minute, though there is no public report of how many traded at the particular price that I received. On average over all the trades selling short mispriced puts, the number of puts traded the same minute as my “fill” on C2 exceeded 30 contracts. The most contracts that I traded (after rescaling) in a single order was 31.12 shares, and the typical trade was much smaller than that.

As I noted in an earlier post, if selling puts bothers you, you might take a look at the two periods where I did no naked sales of mispriced puts. In those periods on C2, I did 0.85% a day, which works out to an 8-fold return over 252 trading days in a year.

Obviously, such a high 8-fold annual return would far exceed the results of my backtesting, and thus should not be expected going forward. But these two periods reflect trading mainly the main timing model, with discretion used to determine the size and precise makeup of the directional position and any hedging to be used.

13

If you truly believe that you:

prove it by trading futures here on C2. Option trading (which I have 30 years experience of) is notorious for slippage, deceptive fills, innabilty to fill, unable to place effective stop-losses, and a myriad of other practical difficulties that only show up if you TOS in actual trading. I’ve also had excellent “models”, but could never implement them in actual real life options trading.

2 Likes
14

I’m not sure what you mean by stalled quotes, but if you mean that some reporting system showed some stale quotes at a price at which there were no actual trades, just stale bid prices that no one could actually trade at, that’s not what happened.

Remember, none of my opening trades are fills on a bid or an ask, All of them are trades at prices where real options were bought and sold on an options exchange after I placed my limit order, mostly hours after I placed my order.

And all of them met one of the criteria that you expressed as aspirational in your email to Matthew Klein that he quoted: the options traded well past the price on my limit order. IE, they “traded through” my limit price.

For example, on March 31, I placed a limit order to STO UVXY1707P15.5 at limit 1.57. My trade was “filled” by C2 when options actually traded at $2.92. 176 options at that strike and that expiration traded that minute, though it’s not public how many traded at that high price.

15

I am sure everyone will appreciate an answer to this question.

16

I noticed an initial capital of 2000$ in your model. Since you have disclosed You are an academic I must assume you know exactly what that it (should) mean. Is that the amount suggested to your potential subscriber to start trade your strategy or as I suspect is an amount kept insanely low to just boost insanely your performance?

I agree with Elemental here, You seem knowledgeable but the trick used, I am sorry is a big red flag including the private strategy kept private during incubation. 2000$ is an amount easy to burn.

Interestingly you are the second Academic that take advantage in a big way of C2 holes…

17

  1. Most systems here are not TOS. I believe that among the 12 most profitable systems on the Leader Board, only one is TOS. I’m a bit surprised that some here appear to think it’s important whether I go TOS.

  2. In a regression analysis analyzing what predicts future performance at C2 (some of the results of which I discussed here in the forum), I discovered that the TOS variable did not predict higher returns. Though the period of returns analyzed was too short to be at all conclusive, it made me think that TOS might be mainly a marketing ploy that makes prospective investors more comfortable, but does not actually protect them.

  3. If I get lots of subscriptions continuing over a several month period, I will probably become TOS, but I want to be reasonably certain that the extra cost and hassle is worth it in additional subscriptions. Does it matter that much to prospective subscribers? Most successful developers here do without it,

  4. I have a very large amount invested elsewhere in my basic XIV timing model. I would be shocked if any individual invests anything close to that in my strategy on C2, but I will see.

1 Like
18

[quote=“QuantitativeModels, post:17, topic:9830, full:true”]
Most systems here are not TOS. I believe that among the 12 most profitable systems on the Leader Board, only one is TOS. [/quote]

  • Most systems here are not worth subscribing to, even for free.
  • As well as most systems on the leader board.
  • Profitable over a certain period means very little when there is a large tail risk or others.

Point is:

  • If you don’t risk your own money on your strategy, why should others risk theirs?

Excellent, so the way to prove this claim is to go TOS.

2 Likes
19

Wow. I think most of your comments are uncalled for.

First, do you understand how returns are figured here? I started with $25,000. but as the portfolio got bigger I rescaled several times. The purpose was to make it easier for smaller investors to invest in the strategy. The effect of rescaling lower is to LOWER the return, not increase it. The commission and autotrading fees are applied to the small portfolio and hit them hard, much harder than larger portfolios. If I had wanted to get higher returns, I would have left it where it started, and it would now be over $500,000.

But I had read the many posts on the forum lamenting that strategies with large portfolios were too high to make scaling easier for moderate investors, and I thought I would be sensitive to their concerns, even though it reduced my reported returns. For being considerate, I get abuse from you based on a misunderstanding of how returns are figured here.

BTW, I recommend a minimum of $10,000 to start.

Second, people on the forums also complained about people going public too soon and soliciting subscribers when they had a few days or weeks of a track record. When I said that I was going wait longer than most to go public, I got the impression that people thought it was a GOOD idea. Now you try to make it seem somehow a bad thing. I actually considered waiting 6 months out of deference to investors. Imagine how horrible you would have thought I was then!

Look, I could have started earning subscriptions much earlier than I did. I was urged by another of the developers in private messages to charge a multiple of what I am charging. I am not being philanthropic here, but I think I am offering a good deal to potential subscribers, the lowest sub price ($77) among the 20 highest return strategies on the Leader Board. I don’t expect most potential subscribers to jump at the opportunity; anyone should rightly be skeptical. I don’t think I’d invest in a stranger’s strategy after only 82 days. But I would follow it starting now.

On this forum, I have urged C2 to add a column to the Grid and the Leader Board for (total) returns since autotrading began. If I were investing here, that is what I’d look at most strongly, along with DD and strategy age.

There are many other good volatility strategies here at C2; feel free to go to them if it makes you happier.

But most of all, relax. Time will tell.

1 Like
20

Many people here like TOS strategies, but if you do a research on TOS strategies, you will find:

  1. In leader board top 20 or top 50 strategies, TOS strategies have larger drawdown than other strategies with the same level of returns;
    2.In leader board top 20 or top 50 strategies, TOS strategies have lower return than other strategies with the same level of drawdown;

Why this happens?
Because most C2 strategies are designed to attract subscribers than to make money itself.
So most C2 strategies are skewed towards higher return/DD ratio, because only this type of strategies are most welcome by subscribers.While real traders don’t care much on that . I mean real traders care more about higher return than lower DD.
Real traders use Kelly’s criterion, or 1/2 Kelly’s criterion.
A typical real trader strategy is like YM Magic, with high leverage, high DD and high return.
I myself use much higher leverage in my real account trading, which would not be accepted by C2 subscribers, that is why I don’t go TOS.
By real traders I mean individual trader, who trade for themselves.
Funds traders who trade for other people’s money do care more about lower DD( because they also need to attract clients).

So why C2 subscribers care more about lower DD?
Now if I trade for myself, my account goes like this: $10000-> $30000->$12000
I have a 60% drawdown but I can tolerate it because at that point I am still up $2000.
Now for a strategy at C2, there are subscribers who join at the time when the account is $30000, for those subscribers, when their money go from $30000 to $12000, it is not accepted and they can not tolerate. They are very likely to unsubscribe when the drawdown is 40% or 30% or lower, and never come back. So to pick a lower DD strategy is extremely important to them.
This is why individual traders accounts can have higher DD than C2 strategies and can still survive, and you can’t say if you are not TOS, you are not to be trusted.

6 Likes