Show HN: Find the 10 highest and 10 lowest correlations to any stock (betagainst.fun)
During the start of the year I was thinking how could I bet against certain stocks (in my case mainly Tesla) without using derivatives and the risks that come with them.
After I had success betting on the oil price with a highly correlated investment fond, I came to the conclusion that negative correlations could be used to bet against the price of other assets. Unfortunately, it is not easy to find correlations between assets if you don't know which assets to compare in the first place.
So I created a website where you can find the 10 highest and 10 lowest correlations of certain assets.
146 comments
[ 4.7 ms ] story [ 34.7 ms ] threadCool idea, by the way. I'll have to play with this more.
https://betagainst.fun/asset/_gspc_spy/
https://betagainst.fun/asset/_dji_dia/
You cannot compare the same or roughly the same basket of assets against itself.
It's a concept that fits nicely in the category of technical analysis, but not one that allows you to find any particular market insights.
If you took the whole tradable US securities market and said "what doesn't correlate with these price movements?" it doesn't mean anything.
If it did, I would ask you, "What explicitly do you _think_ this means?" Just because you get numbers in and out doesn't mean you're working with anything meaningful.
The only broadly meaningful concept you can get out of anti correlation with the market would be bonds because interest rates increasing push down the estimated future cash flows of publicly traded companies, affecting their valuations.
Everything else can be considered speculation.
It's useful for hedging risk
> The only broadly meaningful concept you can get out of anti correlation with the market would be bonds because interest rates increasing push down the estimated future cash flows of publicly traded companies, affecting their valuations.
This is wrong. Consider pair trading or long/short strategies, both of which rely on estimating correlations.
"Beta is a measure used in fundamental analysis to determine the volatility of an asset or portfolio in relation to the overall market."
"Technical analysis is a trading discipline employed to evaluate investments and identify trading opportunities in price trends and patterns seen on charts."
That's just looking at pricing. It's the same category as technical analysis.
If it did, I would ask you, "What explicitly do you _think_ this means?"
Caveat: I'm an idiot and just trying to understand from a layman's perspective.
Wouldn't you be able to use this to invest when you think the market, as a whole, is overvalued?
For example, if the tool showed that a 30-year treasury bond (or similar) was negatively correlated with the S&P 500, wouldn't it suggest it was a good idea to buy bonds (or similar) when I thought the overall equity market was overheated? The idea being there are certain industry/stocks like maybe precious metals/mining that do well when the rest of the market is tanking?
It's literally asking "How do these roughly 500 plus or minus largest stocks correlate with the largely the same basket of securities?"
If you took one company and compared it against the total US market, then you have something at least. If you take the market itself and compare it against itself, it makes no sense.
This is a common metric referred to as market beta.
> If you take the market itself and compare it against itself, it makes no sense.
S&P 500 and Russell 2000 are both broad market indices, but they perform differently. Even using the same index constituents with different weightings (e.g. equal vs cap weighted) can produce meaningfully different results. It makes plenty of sense to compare them.
No one. No one is seriously looking at the Russell 3000 and comparing it to the Wilshire 5000.
To the child comment:
I'm talking about two total market indices, apples to apples, and you're talking about apples to oranges. NASDAQ is a stock exchange, not an index.
Further still, even if you were talking about the NASDAQ composite, they're two entirely different indices. They do not have the same goals. There might be some relative meaning there.
There isn't much meaning between comparing two or more indices that have the same goals and constituents. You're only tracking the differences between constituents at that point, and you wouldn't need beta to do that. You could just calculate the difference between the constituents that are not a part of the total set.
People compare indices all the time to assess relative performance (e.g. Russell to NASDAQ) or HY credit to IG.
Response to edits:
> I'm talking about two total market indices, apples to apples, and you're talking about apples to oranges. NASDAQ is a stock exchange, not an index.
NASDAQ composite is one of the most commonly referenced indices. There's also NASDAQ 100, on which one of the largest ETFs in the world (QQQ) is based.
> Further still, even if you were talking about the NASDAQ composite, they're two entirely different indices. They do not have the same goals.
That's why people compare indices -- they're interested in evaluating different aspects of the market.
> There isn't much meaning between comparing two or more indices that have the same goals and constituents. You're only tracking the differences between constituents at that point, and you wouldn't need beta to do that. You could just calculate the difference between the constituents that are not a part of the total set.
Weightings? An equal weighted index will reflect increased contribution from smaller companies versus a market cap weighting (compare ETFs SPY/RSP). Same constituents, different emphasis.
Also, roboadvisors like Wealthfront trade strategies based on negative correlations within and between market indexes.
Regrettably, I shouldn't have bothered replying at all, some HN readers think beta is a fundamental measure. It is not. Many sources get this wrong for some reason.
If you are looking at price movements, you are doing technical, not fundamental analysis.
Also 80% by some metrics, much, much smaller by others.
That said, in any set of 20+ variables there will be a 10 highest/10 lowest correlating.
Without a good (specific, hard to vary) explanation as to why the correlation happens, I would not use this information to gamble real money.
If you track this over time, you will see that these change (often at times when you really want them not to).
Past observances aren't super helpful in predicting fractal futures.
It's better to correlate daily returns than daily prices, since the latter are nonstationary, and I suggest using 1 year of daily returns rather than 20 since correlations do change over time. When I worked as a financial quant no one looked at 20 year correlations to measure near-term risk.
As such the holding period might be on the order of 1 year for some people, so just 1 year of daily returns might invite too much trading if you're rebalancing.
Not in taxable accounts where short term capital gains are taxed at ordinary income tax rates (or at an even higher rate in Massachusetts). It's difficult to beat index funds on a pre-tax basis. On an after-tax basis it is almost impossible with discretionary stock picking.
Where are you getting the historical data? There's a ton of fun stuff you can do with this kind of long-horizon of data
However, I get a “504 Gateway Time-out” error for https://betagainst.fun/asset/bz__f_bno/. HN hug of death?
I thought the opposite of buying a single stock (Alpha) would be in service of.... Beta Gains T.
With few exceptions, all stocks are correlated to the broad market.
https://en.wikipedia.org/wiki/Beta_(finance)
One explanation that would be popular right now is that the price of stocks can be calculated based on expected future cash flows and interest rates (e.g. think of what you'd pay for a bond that produces the same csh flow.) Since the interest rate is a factor in that calculation for all stocks, there's a correlation right there.
The exceptions tend to be sketchy. At one time it was thought that gold mining stocks would move against the market, but what I heard was that gold mining firms are badly run and if you like gold you should just buy gold.
https://en.wikipedia.org/wiki/Pairs_trade
The most famous pair trade was this: in the morning of a trading day you buy the stocks that performed the worst yesterday and short the ones that performed the best. Because the market reverts to the mean, this strategy made money hand over fist from the 1980s when Morgan Stanley started using it until the summer of 2007 when somebody fat fingered a button, unwinded their position during the day, and crashed everybody else who was in this trade.
The profitability of that trade went downhill long before that crash but since then it's been impossible to make money doing it. (When there's a hedge fund bubble the profitability of a trade drops gradually as more people pile into it and the market inefficiency it targets is dispelled, quite different from an equity bubble which goes up dramatically until...)
[1] https://papers.ssrn.com/sol3/papers.cfm?abstract_id=525282
The idea is not to have perfect negative correlations, but to find values with some negative correlation. This way you can profit from falling prices of an asset. Not with a 1 to 1 yield, or even higher.
If your goal is to manage the maximum loss from a short position, it’s much easier to short the stock and buy a protective put. This gives you perfect exposure to your bet, while allowing you to manage your loss exactly.
> Not with options.
you maybe meant to say "not when you buy options":
> 2. Sell a Call. This gives someone the right to buy the stock at a certain price (and you promise to sell at that price). If the stock is below that price your profit is what you sold the call. Your maximum loss is infinite since a stock price has no (theoretical) limit.
the entire point of everything under discussion here is that "betting against a stock" can take multiple forms. if you'd like to pick one, google, or your interlocuters, could tell you what the maximum possible risk is.
keeping your money in cash is a form of betting against a whole lot of stocks simultaneously.
But put options? Surely all you can do is lose what you bet in the first place?
The value of a call option is unbounded.
This link[1] has some good questions and answers about options, although it certainly isn’t complete (e.g. dividends are mentioned but voting rights are not).
[1] https://www.blackwellpublishing.com/content/kolb5thedition/c...
[2] https://news.bloombergtax.com/tax-insights-and-commentary/ma...
I got myself mixed up and was misled by a Corporate Finance Institute quote (which incorrectly notes the loss is unlimited only to contradict itself):
For the seller of a put option, things are reversed. Their potential profit is limited to the premium received for writing the put. Their potential loss is unlimited – equal to the amount by which the market price is below the option strike price, times the number of options sold.
Also, prices for these options (Tesla is a good example) aren't cheap. For example a bet that Tesla will be below $900 by Jan 19th 2024 will cost you about $250 / share at the moment. That means Tesla needs to actually be below $650 on Jan 19th 2024 before you make 1 dollar.
As long as you understand whats going on and what you are actually betting on, you are fine. But options have lots of 'gotchas'.
https://en.wikipedia.org/wiki/Put–call_parity
i.e. When you short a vanilla equity, you'll likely have transaction cost of borrowing the stock as an interest rate over time. This cost is incurred as you keep the position open. This cost is related to the cost of capital for the shares you've borrowed. Cost of capital is an implicit cost on time.
To construct a synthetic short (using options/bonds), you basically short a synthetic equity. A synthetic equity can be constructed through a long call position, short put position, and long bond position. This synthetic will mostly replicate the the stock's return. To turn it into a synthetic short, you just do the reverse, short a call, long a put, and short a bond.
A bit like Negative Chess...
Leverage + Interest Rate Risk + Liquidity Risk + Settlement + Delivery Risk + Operational Risk + Volatility Risks + Unregulated By Governments And/Or Central Banks + Risks of Error and Hacking + Human Error...
All together! Remember, we are trying to lose money as fast as possible...
But unlike selling stuff you don't have (shares, options etc), you won't lose more than your stake. You have $1k, you put it on red, you invest in the S&P, you buy Roubles, you buy a meme stock, you buy a TSLA call for 6 months time at $2k, you buy dogecoin, at most you will lose $1k.
You borrow 100 shares of TSLA when it's $900 and sell them, you'll get $90k now, but you might end up with TSLA jumping to $5k overnight and you have to send those shares back, that will cost you $500k, well done.
The bigger risk comes in selling options: you sell a call and (assuming you don't already have the underlying shares) your downside is unlimited - you might have to pay sky high market prices only to sell (to the call option holder) low; you sell a put and your downside could be the entire strike price - the underlying could have gone to zero but you're forced to buy high (at the agreed price).
To me I don't care if a certain stock is correlated by something, I would more like to know which stocks do have correlations or if there are correlations with a time lag
Of course, the problem with something like this is that using it would, of course, ultimately change the market so that x→0.
These are starting to look like perpetual motion submissions to the patent office.
- Building things like this is always great. And its a fun site to poke around on.
- I would not count on this approach or expect it to be reliable in terms of actually hedging. Correlation, as a measure, has lots of issues. You are boiling down a lot of complex relationships into a single number. While it is convenient for many calculations, there are many problems. For example, many asset classes will go through periods with positive correlation and then later, negative correlation. This is due to a factor driving both securities price becoming more or less volatile compared to the other drivers. E.g., recent increased volatility around inflation expectations driving correlations between rates and equities. Whereas, few years ago, inflation was not driving anything.
- One alterative approach is to have a "risk model". Which essentially decomposes a security into drivers. Each security then represents a basket of these drivers. You can then use this model for range of purposes. While not perfect by any means, the model contains more information than a correlation. These too have a range of issues and creation and use is as much art as science.
- In general, you won't find many negative (or even very low) correlations across individual equities. Most stocks are driven by a common set shared risk factors that drive much of the risk. But if you can find negatively correlated securities (or lowly correlated), then that is certainly helpful.
And most people blow up their account, because their trading strategies worked for a long time.
Speaking for a friend, of course.
Setting stop loss for qualified trader should be like looking around when you are crossing the street on intersection.
There some people may have gambling problem, or software has bugs, but it is another story..
I've never understood stop-loss, it seems to me that by using stop-loss, one is basically saying "I don't know what the fundamental value is and if it goes down I want to sell". Then, why buy the thing in the first place?
A strategy that indiscriminately buys more of an asset as the asset price goes down only works with an infinite bankroll. Such a strategy run with a finite bankroll will necessarily blow up eventually. Rigid stop losses mostly just increase the volatility of a strategy though (which also contributes to portfolio drag). A sustainable strategy harvests volatility of the underlying (mean reversion) and/or dynamically deleverages (momentum) as the price moves against the entry.
because your strategy/estimation looks like not perfect in this case, and you may consider to cut your losses at 38, and not become broken if it goes down to 20.
Trading strategies is more for shorter term investors, when circumstances can change very fast while you are visiting bathroom, and induce significant losses, that's why you need stop loss to limit your risks in such cases.
Having said that, if you can find a negative correlation AND there is a strong reason to believe that the negative correlation will hold as long as X and Y hold, then the tool could be very useful.
The TLDR is more that correlations throw a lot of information away.
The correlation is only a first indicator.
Is it more likely to find negatively correlated securities across industries, like the SPDR ETFs, rather than individual stocks?
One small typo: Highest -> Higest on the graphs