QuantConnect Community

Just something to check with our European friends, Primoz Golle, Dimitar Dimitrov and Davidii111: I understand that trading US-based ETFs is not possible due to regulations. I just wanted to check: Could you trade call options, e.g. on TLT, instead of trading the actual ETF (assuming you have successfully requested the trading permission for options on IB)?
For instance, here is a Yahoo Finance list of options for TLT: list. Could you for instance trade the first option in the list, TLT210604C00110000?
If yes, it might just be a question of creating a version where we do not trade TLT but a corresponding call option, or a basket of suitable call options.

If options are not a possibility, wouldn't it be possible to trade an EU equivalent ETF if running LEAN as a local instance? e.g IDTL

When running live, the algo intermittently does not pick stocks when starting algo or going back in. Just had in signal Monday and no stocks were picked again. Starting the algo 5 minutes after open instead of exactly at open just in case launching for the first time and warm up needs to happen. Anyone else encountering this?

Thanks for sharing the issue, Ton86. I am wondering whether adding “not self.Portfolio.Invested” as an additional condition for refreshing the universe and rebalancing the ‘in’ and ‘out’ side might help address the issue.

Just one quick additional thought, Ton86: Not sure whether you consistently use minute resolution for all algo components. Using minute resolution might also help address issues in Live trading, see the earlier post by Alexandre Catarino.
Let me/us know how you go. When you have figured out what is going (or someone else does), I think that these will definitely be interesting lessons learned for many, so feel free posting these here in this thread.

Thank you Peter Guenther. I have been running “res = Resolution.Minute” since going live. I will try adding “not self.Portfolio.Invested” next week as an additional condition and provide feedback. Best.

Hi, My understanding is that there are two variants:

a) SEL[QualUp] + I/O[In & Out]

b) SEL[Earnings Rockets] + I/O[The Distilled Bear] 

does anyone have the correlation between the two strategies. If the correlation is low then it might be worthwhile to run both in parallel.

That is a valid thought, Narendra Kulkarni!
Technically, there are more possible combinations, which we get by mapping the full factorial of the stock selection strategies that were discussed in this thread ([TQQQ], [ValuationRockets], [QualUp], and [F-Score]) and the in & out strategies discussed here ([InOut] and [DistilledBear]). So, this gives you 4 x 2 = 8 possible combinations.

In terms of the correlation idea, what you could be doing is to run the backtests for the two combos you have listed and, in the backtest results (you may have these already available from your earlier tests), copy the monthly returns (select: Total Net Profit) from the results' Overview tab into an Excel sheet to calculate the correlation using Excel's correl function. If you are happy to share, it would be interesting to see what comes out of this calculation.

Thanks you Peter. I can do what you suggest, however I would prefer to do using code. Is there any way to compute strategy correlations using quantconnect code? If someone can give me any pointers, I am happy to code. Secondly, my thought was that if the correlations between the different variants are low enough, we could simply combine them using some formula. Maybe risk parity, or equal weighting. More importantly if the strategies are overfit, hopefully these errors will somewhat cancel out and we wont get burnt in live trading.

My understanding of the quantconnect platform is that for us to run multiple algos in parallel we need to use the multiple alpha models. So should we first focus on converting some of the existing code to the multiple alpha framework? This will allows us to combine the different versions of in and out and run them in parallel.

Fantastic idea, Narendra Kulkarni! I remember that a few months ago Zicai Feng has translated Strongs's Piotrosky F-score algo version into an Algo Framework (link to the post). So this might be the perfect base to build the Multi-Algo Framework from.

The suggested quick-and-dirty correlation test using the Backtest returns of the individual strategies could be an initial face-value assessment regarding the likely gains of combining the strategies in a multi-algo framework. However, the multi-algo framework will provide a more precise picture.

Implementation re Live trading recovery
vidal boreal correctly noted elsewhere that the self.dcount and self.outday variables reset to zero when live trading is interrupted, meaning that the algo effectively ‘forgets’ that it went out of the market and wanted to stay out for some time.
The attached backtest uses the ObjectStore to save the dcount and outday as important state information of the algorithm that can be recovered in case of an interruption. To illustrate the approach, I am using the old (p. 2 of this thread) [In & Out / Distilled Bear] + [QualUp] code. The important lines are:

Lines 298-301: Tries to retrieve dcount and outday from the ObjectStore if they are zero (i.e. the algo restarted) and the info exists in the ObjectStore.

Line 334: If running live, a function to save dcount and outday is called on a daily basis.

Lines 416-418: This is the function that actually saves dcount and outday to the ObjectStore.

Feel free to copy and paste these lines into your algo version.

hmm… I am receiving the following error stated below when I ran the algo attached in the post: :Implementation re Live trading recovery. I remember solving something like this before. Let me see if I can find a solution for this.

Thanks for sharing this bug, spacetime. Second try. I have made the following additions/changes:
1. Line 297: check for whether self.history actually has data. Only proceed if it does.
2. Line 245-246: saved the benchmarks for the debt and tips basis differently, without requiring that dcount is 0.
3. Line 255-258: removed a bunch of double squared brackets which are not required.

Let me know in case the bug persists and I'll have another look.

tested it from self.SetStartDate(2015, 1, 1) and have not received any errors

Disregard…Cloned the wrong one

Peter Guenther I'm trying your last posted Algo and getting a runtime error, any idea why?

Runtime Error: Trying to retrieve an element from a collection using a key that does not exist in that collection throws a KeyError exception. To prevent the exception, ensure that the SHY key exist in the collection and/or that collection is not empty.

  at _validate_read_indexer

    key=key in indexing.py: line 1176

  

Given the current equity market uncertainties (esp. regarding inflation expectations driving up bond yields), this might not be the best time to make risky leveraged investments in Tech. Anyway, I could not help it and had to try out the new In & Out v8 algo with one of the ‘old classics’ of this thread: holding the TQQQ :)

Somewhat relevant for the TQQQ holding strategy in the previous post and discussed earlier in this thread, I found this paper on SSRN (by Lewis A. Glenn). Title and abstract below.

Title: Long-Term Investing in Triple Leveraged Exchange Traded Funds
Abstract:
The phrase long-term investing in triple leveraged ETFs is somewhat of an anathema for investment academicians. As a result of daily rebalancing and so-called beta slippage or “the constant leverage trap” it is highly likely over the long term to significantly deviate from the targeted leverage and, in so doing, generate wipeout losses. In fact, the data show that many 3X leveraged ETFs have performed poorly over the long term in the period of their existence. However, there are some significant exceptions. The 3X leveraged Nasdaq ETF, TQQQ, for example has delivered a total return exceeding 10,000% (CAGR of 54.4%) over the 10+ years of its existence, more than 13 times the total return of its underlying ETF (QQQ). Similar, but less impressive numbers have been achieved with triple leveraged ETFs that target the S&P 500 and the Dow Industrials.

These results, however, are accompanied by excessive volatility and very large maximum end-of-month (EOM) draw downs, in excess of 49% for TQQQ, and with maximum daily drawdown of almost 70%. Many retail investors would be unwilling to sustain such (paper) losses. Here we propose a strategy to mitigate these large draw downs while still retaining much of the upside performance of triple leveraging. We show that by simply establishing a portfolio consisting of an equal dollar amount of TQQQ and TMF (the triple leveraged ETF linked to the 20+ year treasury bond), and performing bimonthly rebalancing, a total return, after transaction costs, in excess of 5,800% (CAGR of 44.9%) would have been achieved over the 10+ years of TQQQ existence, with the maximum EOM draw down less than 25%.

Hello everyone. 

I'm pretty much a beginner with QuantConnect (just started to vaguely understand how to create and use a Dynamic Universe). I have read through this thread and just wanted to say it is awesome, I hope I can get to the level of detail and complexity that you lads are dealing in at some point. 

Peter Guenther, is the most recent version of the algo you posted updated with the changes you made for “Implementation re Live Trading Recovery?” I tried having a look but in all honesty got a bit lost.

Also, is this version suitable for live trading. Alexandre Catarino gave a couple recommendations, including to use Resolution,Minute. Has anyone tried it out live and if so, does the algo make the trades we would expect it to make. I am happy to try this out myself but I fear I would not have the know-how to really grasp whether it was working or not, or suggest any changes.

Forgive my ignorance :)

No worries, JSO 2045, thanks for joining the discussion!

Regarding your questions:

(1) Does the algo have Live Trading recovery?
It does. The key variable for the in & out decision is the density of the number of signals saying out (see the black line discussed and illustrated in this post; the corresponding variable in the code is self.signal_dens). In Live Trading, this variable is saved every day that the code runs to the ObjectStore after the variable has been updated. See line 139 (if self.LiveMode: self.SaveData()) which calls the function SaveData that saves self.signal_dens. For the SaveData function, see lines 190-191 in the code. On the next day, if Live Trading, the code loads the saved variable from the ObjectStore before running its other daily checks and routines. See lines 98-102. So, if the code get interrupted, it would not matter much since it would pull the most recent critical information from the ObjectStore before it does anything else. This means that it has a recovery mechanism in place for live trading interruptions.

(2) Does the algo use Resolution.Minute?
It does. See line 27, where the variable res is defined which is then used for all equity tickers that are added (e.g. in line 31).

(3) How would I assess whether the algo performs the trades that it's supposed to perform?
One approach could be to run it live for a certain period of time and then run a backtest for the same period of time, comparing the trades.

Hi all, there's been some discussion on possibly using options with this algo.  I've modified the algo to also use options when out of market.  you can set self.useOptions to True or False to compare.  Thus far using put options as a hedge does not result in a better return, but I have not done any extensive testing to find more optimal parameters for using options.

Thoughts…
1. What is the optimal DTE for the options.  Perhaps we could look at the avg time out of the market for this, or maybe it makes more sense to use a short DTE and keep rolling while we are in out of market mode.
2. When is the best time to roll?  Right now I am just using 60DTE and rolling at 30.
3. What is the optimal weighting to devote to the options portion?  Currently whatever is not devoted to options is devoted to the normal out holding (TMF, TLT, etc).  I tested with 1%, but perhaps this is too low or too high.
4. How far out of the money should we be purchasing put options on the underlying?  I tested with 20%, but perhaps it's better to be closer to ATM.  I think further OTM is probably better for black swan protection as we can acquire more contracts for the money we are devoting to the options strat, but if the algo is truly good at taking us out of the market at optimal times, then being ATM might be better.
5. There are some very large drawdowns, due to having large gains in the options that are erased before the options are sold.  You will see this much more if you devote a larger weighting to the options hedge.  Currently the options are sold only if the algo goes back in the market, or we hit the time to exit, (it sells the options and then buys new options if the algo is still out of market).  It might be a good idea to create some sort of profit taking strategy to sell the options if they have increased by some amount, or to trigger a trailing stop after they have increased by some amount.

NOTE: do not use QQQ as the underlying for the options.  There is an issue with the prices starting in August 19, 2015.  This will make the algo look insanely profitable, but it's really that the options prices are off by a couple orders of magnitude. 

Peter Guenther For 1) That is pretty clever, thank you for explaining it for me. I'm going to have a look at all of it again and try to understand as much of the code as possible. I saw that there is another discussion going on currently about people taking the In and Out strategy over from Quantopian so i might give that a read as well to try catch some more info.

For 3) I think I am going to take your advice and try it live to see if it works (as well as hopefully making some decent returns as well). The only thing i am worried about is that there is some inflationary pressure at the moment which might not translate into great returns. 

Thanks for sharing this options implementation and structuring the discussion, Jon Bailey. Nice job. I also just quickly wanted to copy in Elsid Aliaj since I know that Elsid is looking for tail-risk hedging à la Taleb and your out-of-the-money put options implementation is a step into this direction. Of course, to fully implement the tail-risk hedging, the put options would need to be continuously held, i.e. during in and out regimes, but this is only a small modification. I reckon the smaller returns that we observe when holding the puts make sense, since tail-risk hedging is like an insurance and we usually need to be prepared to pay a premium for an insurance.

@peter-guenther , actually most of the code is from a tail risk hedging (Taleb) algo I was working on when I found this thread.  What I found is that keeping the hedge on 100% of the time is a pretty big drag on returns, and then wham, the market tanks and there are massive returns.  Basically the returns you get from the equity holdings are pretty much paying for the hedge and keeping stability of principal.  You are just waiting for a market crash to cash in.

I thought, if only I could better predict when a market crash/correction were to take place.  Of course this is a fools errand, but if I could come up with a way to know that it was mostly safe to be in the market without a hedge in place, I could dramatically increase the returns during the bull market years.  What I invision is some sort of confidence indicator, with three levels.  Above some level we are confident and the markets do no need to have any hedge in place.  Below that level but above the lowest level, we are still “in” the market and long equities, but with a tail risk hedge in place.  And then below some level of confidence, we are “out” of the market and moved into bonds/treasuries, along with the tail risk hedge.

At some point I plan to see if I can adapt the in_out algo to have this confidence indicator and if it even makes sense to do it, or perhaps it is just better to keep the tail risk hedge on all the time.  What I do see, is that there are some periods of significant drawdown while we are still “in”.  The “out” determination is not always timely.  However those may be the periods where the middle confidence level comes into play.

Great ideas, Jon Bailey!

Musing a bit about options how this could be operationalized: 
To specify the regime “in the market but with hedge” one could also use the percentile approach but with a larger alpha, e.g. 10 instead of 5 percent. This would mean that the hedge is put in place earlier (i.e. already for softer signs of stress) than the out decision is being made. For example, one could specify:
activate_hedge = returns_sample.iloc[-1] < np.nanpercentile(returns_sample, 10, axis=0)
and then activate the hedge if: activate_hedge[self.SIGNALS + self.pairlist].any()

In addition, one could try to use a broader set of signals for the “in with hedge” regime, e.g. including the Yen or Swiss Franc to identify safe haven moves, or additional markets to check for stress signs. I will have a look.

@Jon Bailey found two doomsday hedges on the internet with for OTM VIX puts. Just searched but didn’t found the address. the Setup was like this:

Assuming the VIX is around 15, you would sell calls on the VIX on 20 and than buy the double amount of call on around 25 or higher with the total sum earned and spend equals 0. I think they were 60 DTE. Nothing mention how to define level the first call options. Second level was calculated by double amount op options and sell/buy equals zerro. They should expire most of the time with the vix below the sold putts to make it a cost free hedge. The risk would be if they expire with a VIX between 20 and 25. This was setup every month and the options were hold until expiration date. the total risk, was 0,25% of capital hedged on each of this 4 sets. Totally 1% risk based on capital to hedge. (In that case the risk is 25-20= 5 x number of option)

An other concept I found was to just buy very far out of the money options on the VIX. They used VIX 100 calls. They bought 60 DTE and rolled around 21 DTE. They did not mentioned the amount needed to hedge. 
 

I am running the V8 version and noticed it recently switched to TLT and back in December it had switched to stocks just before the tanking of stocks. I know these might be exceptions but just trying to understand and see if anyone knows why?

😊

Sector rotation, first tapering event, and seasonality contributed to algorithm buying behavior and your results.

Seems to be getting whipsawed back and forth at inopportune times. As long as it's providing alpha it shouldn't matter. The question does stand though if it should move into cash and/or possibly commodities instead of bonds during times of fed funds rate increases. Currently the drawdown isn't much better than nasdaq's due to the increase in treasury yields. Should get pretty interesting this year and next as the tapering begins+rate increases.

This comment has proven to be spot on, Anthony B. Thanks for sharing / the early warning!

The strategy seems to fail in the 2022 regime where now 20y bond crashes together with the market. Then it looks like the strategy was overfit on previous regimes or the hypothesis that bond is a safe hedge does not hold up :)

Santa24 The main problem why the strategy didn't took that into account, is that it never “faced” an inflation-shock situation like the current one, where expected inflation is far lower than the realized one.

During these kind of situations, nothing other than commodities outperform the market.

If you would have been backtesting the strategy during the '70s (for instance) you would have spotted this. But free to believe the overfitting reason :P :) 

Hi all, newbie joining the thread here.

What'd the resources to read/learn on how to monitor and which metrics to keep an eye when having an algo trading live. For i.e., if I happened to have this algo running for the last couple years, what should I have looked to preventing it from trading in 2022.

If the questions is a bit off-topic in this thread I am happy to repost it in a proper thread.

By the way, very impressive work Peter Guenther!

Santa24, ingenue question here: what'd be the key pieces you look at to determine an algo is overfit or not? Just looking for learning on how to spot such things. I am coming from TradingView's Pine code, and I found myself in hot water recently due to overfitting my algos in there. I found QC while researching for this problem and I am liking a lot the platform so far.

Rafael Trevisan i am by no means an expert, but the issue I see with the in-out approach is that it does very few trades, holding positions many days and QC has only data since 2002. Now, as pointed out correctly in the discussions, the in-out strategy performs very well 2008-2021, but then takes heavy losses because the macro-economical environment in 2022 is fundementally different from 2000-2021. As Simone Pantaleoni  says: “the in-out has not seen an inflation-shock situation like the current one, where expected inflation is far lower than the realized one”. I can only think of the following remedies:

• backest and research from 1960 - 2002, do out of sample validation 2002-2020 :)
• add create macro-economic rule overlays to the in-out, for example, if (real_inflation - expected_inflation) > threshold, do NOT trade at all or do commodity mean reversion

One might theorize that after the inflation shock, the in-out may start to work well again :)

Rafael Trevisan IMO, this strategy works when US markets are growing and S&P500 grows as well. When markets are in decline it will not work. There is no way to tell if it will recover or not with technical analysis, because this is a political problem, not technical. If politicians will fix it somehow, it will start working again, if not, then no 🤷‍♂️

Rafael Trevisan The general rule of thumb is to backtest on ~70-80% of all of the data you have. Then validate the results out-of-sample on the remaining ~30-20%. If it performs poorly on the out-of-sample then the algorithm is overfitting. But that's not enough. You will also need to fully understand the ins-and-outs of why the algorithm works and what could cause it to fail in the future. e.g when going “out” of market the In-And-Out goes to only bonds assuming that bonds are uncorrelated to the rest of the market and is always a safe haven, but we can see recently due to the inflation rate shock that assumption cannot hold. 

Additionally I see that more and more people are interested into running this strategy ever since Quantopian shut down. But we will have to question how well this strategy will fare as more and more people are running it on scale, would Alpha in the future be exploited away or not?