Armando Gonzalez, CEO, RavenPack
| February 25, 2020
Jack Schwager’s interviews with top traders highlights their decision-making processes - but how could RavenPack data have added value, if any?
Market Wizards is considered a set text for anyone who wants to be a financial trader. Written by a futures analyst, Jack Schwager, it is comprised of 16 interviews with some of the world’s most successful traders, including Richard Dennis (of ‘turtle trader’ fame), Paul Tudor Jones, and Ed Seykota.
The book’s value rests on its unique access to the workings of the minds of this privileged group of top traders who between them averaged returns others could only dream of. Almost all were able to turn initially modest stakes in the four-figure region into multi-million dollar accounts. Re-reading it again, from an alternative data perspective, has yielded up some fresh insights on the value of that data, specifically RavenPack’s news sentiment data.
Critics of alternative data often point to the lack of understanding by vendors of how the industry applies its data. In systematic trading it is pretty clear: sentiment time series can provide unambiguous buy and sell signals which can be incorporated into algorithms; re-reading Market Wizards, however, also provides case studies of how alternative data might fit into a discretionary approach.
Below we take a closer look at 3 of the traders interviewed by Swager, both for the awesome insights they provide about their trading style, and with an eye on how our data could add value to their processes or those similar to them.
For any trader - wizard or not - the impact of the news cannot be underestimated. The first trader Schwager interviews, Michael Marcus, had his best year in 1979, when he traded Gold up from the mid $200s, as it surged parabolically to a peak of over $800/ounce. News of the invasion of Afghanistan stimulated the initial breakout move. Interestingly, Marcus says he heard the news before traders in New York did because he was stationed on the west coast. The advantage proved key in front-running the subsequent rally:
“In those days, I had an advantage by being in California, because I was up trading in Hong Kong when my New York colleagues were asleep. I remember when I heard about the invasion of Afghanistan on the television news. I called Hong Kong to see if anybody knew about it, and nobody seemed to; the price wasn’t changing. I was able to buy 200,000 ounces of gold before anyone knew what was happening,” says Marcus.
Marcus goes on to say it was an opportunistic news-based trade that he never again replicated, suggesting that even in 1979 markets were highly efficient. From a news sentiment-based perspective, this example highlights the value of being ahead of the curve. Here the RavenPack platform’s alert functionality does a great job of keeping trader’s up-to-date with the latest headlines.
The next trader in the book, Bruce Kovner, is a wizard at trading currency markets. What’s really interesting about his interview is the insights it gives into his thought processes. Kovner talks about how he makes decisions by ‘imagining’ different future scenarios and then whittling them down to a central scenario by a process of deduction.
“One of the jobs of a good trader is to imagine alternative scenarios. I try to form many different mental pictures of what the world should be like and wait for one of them to be confirmed. Inevitably most of these pictures will turn out to be wrong... but then all of a sudden you will find in one picture, nine out of ten elements click. That scenario becomes your image of the world reality,” says Kovner.
He then goes on to give an example involving the Dollar in the aftermath of the 1987 October 19 crash - the question was would it rise or fall? One scenario was of the Dollar rallying due to its role as a safe haven whilst another was that it would weaken if the Federal Reserve decided to step in with emergency monetary stimulus - either was possible but which would it be?
“It was then that it all coalesced in mind. It became absolutely clear to me that given the combination of a need for stimulative action, dictated by tremendous worldwide financial panic, the reluctance of the Bank of Japan and German Bundesbank to adopt potentially inflationary measures, and the continuing wide U.S. trade deficits, the only solution was for Treasury Secretary Baker to let the Dollar go. Someone had to play a stimulative role, and that someone would be the United States,” says Kovner.
In another example, he likens the process of alternative scenario building, to trying to ‘step into the shoes’ of the New Zealand finance minister and imagining what policies he would be most likely to enact.
“It is trying to figure out the problems which the finance minister of New Zealand faces and he might try to solve them. A lot of people will think it sounds ridiculously exotic. But to me, it isn’t exotic at all,” says the trader.
How could alternative data help in this ‘scenario building’ exercise? The answer is in much the same way as it can in forming any future market view - by increasing the information the trader has at his disposal with which to discriminate between competing scenarios. In the example of the Dollar, this might be by using natural language processing to parse the statements of Federal Reserve members following the crash, as well as informed media commentators, and thereby highlight the underlying drift towards a looser monetary consensus. Indeed
using natural language processing has shown it can be very accurate at predicting future policy moves than established methods such as economists surveys. In the case of the New Zealand finance minister, the evidence suggests news sentiment can provide accurate early estimates of infrequent official macro-economic releases such as quarterly GDP figures, unemployment, or balance of payments data, suggesting it can provide an added piece in the jigsaw puzzle of variables the finance minister might base his future decisions on.
Something else which jumps out of the book on re-reading is how many market wizards tend to use technical analysis - by which we mean price action - to make entry and exit decisions. Here too sentiment data derived from news can provide additional value. Study after
, in most major asset classes, shows a latency gap between RavenPack’s news sentiment data and what data scientists call ‘momentum’, which is really just a fancy name for ‘price’. Obviously, this is the scientific basis for the value of the data, and though an obvious point, it is often overlooked.
The interview with Paul Tudor Jones reveals how he has made a reputation out of one of the most difficult aspects of the game: picking market tops and bottoms. This is to a large degree a technician’s game. Jones is famed for successfully picking and making a fortune shorting the 1986 crash.
“We had been expecting a major stock market collapse since mid-1986 and had contingency plans drawn up because of the possibility we foresaw for a financial meltdown. When we came in on Monday, October 19, we knew that the market was going to crash that day,” says Jones.
Indicators of sentiment have always traditionally been considered a key metric at forecasting market tops and bottoms. These range from the put/call ratio, to number of stocks making new highs - but what about news sentiment, could that also be of value? Research using RavenPack indicates there is the basis for suggesting a broad news sentiment-based gauge for accessing tops and bottoms could be constructed.
showing the data’s efficacy at screening for overbought candidates that are at high risk of reversing trend and falling, for example. Those stocks deemed over-valued - as is the case with the majority of stocks at market tops - when further filtered for negative sentiment were much more likely to tumble than stocks that were similarly over-valued but did not exhibit negative news sentiment. It doesn’t take a huge leap to imagine how using the same principle across a broad constellation of stocks could provide a similar signal for the stock market as a whole.
The figure below shows an idealized version of how sentiment can provide early indications of reversals from market bottoms, from a study called
"Sentiment Reversals as buy Signals"
, by John Kittrell, that used RavenPack sentiment data.
Traders can easily tackle Alternative Data with the RavenPack Analytics Platform, which includes sentiment data on over 250,000 individual entities and 6,800+ market-moving events, available on visual dashboards or via web APIs. Request a trial today.
Please use your business email. If you don't have one, please email us at firstname.lastname@example.org.
We will process your personal data with the purpose of managing your personal account on
RavenPack and offering our services. You can exercise your rights of access, rectification,
erasure, restriction of processing, data portability and objection by emailing us at email@example.com. For more information, you can
Your request has been recorded and a team member will be in touch soon.
High inflation has returned in developed markets after decades of lying low. In our latest paper, we show how to build an inflation-based asset allocation strategy using sentiment data and we illustrate that sentiment-based strategies outperform models that depend merely on past observed inflation values.
This year's RavenPack Research Symposium brought two intense days of knowledge sharing in London and New York, from 25 top experts in natural language processing, quantitative investing and machine learning. Together, we explored how firms can leverage new language models to generate alpha, better manage risk and respond to calls for more sustainable investment practices.
Human capital is at the heart of value creation. Our latest research demonstrates how unprecedented workforce insights, sourced from over 200 million job postings, can generate more alpha.