• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Fleckenstein's views on algo trading

Joined
2/7/08
Messages
3,261
Points
123
I've been following Bill Fleckenstein's column at MSN off and on for almost a decade. He used to make periodic appearances on some television money programs about a decade back and he struck me then as an honest and sensible bloke. He's not a quant. But his opinion on algo trading has hit the mark (I think):

The connection to the quant universe is that Renaissance Technologies, among the biggest quant hedge funds and certainly a very successful one, is the fourth-largest shareholder in both Cornerstone funds.


You have to scratch your head and ask: What is a quant fund doing paying a huge premium for an easily replicated portfolio?

The only logical answer would be that the stock-price characteristics have behaved in a way that makes Renaissance's computer -- which was obviously programmed by someone -- think these funds are a good thing to buy, regardless of the fact that their valuation is beyond absurd.

Meanwhile, a well-placed friend in the quant world pointed out that on any given day, 50% to 70% of stock trading is probably done using a quant strategy of some form. He suggested that folks should think about stocks as financial instruments, looking at volatility, correlation to other stocks, membership in an index and other such characteristics that pertain only to price action.

That's what the computer-driven models at quantitative funds do, setting aside the fundamental questions of what a company actually makes or does and what that business is really worth.

It seems that algo traders are competing against other algo traders in what is in essence a zero-sum game, and where the market or company fundamentals play no part at all. And yet, people like Warren Buffett became massively rich by being "value investors": by carefully scrutinising company and market sector funadmentals before settling in for the long haul (and being impervious to short-term fluctuations).
 
Buffet became massively rich well before the whole algo trading thing came to the picture.
 
I read the article and it seems there is nothing with substance in there.
 
I read the article and it seems there is nothing with substance in there.

Is there anything substantial to say about algo trading except that it has nothing to do with the inherent soundness of the company and/or the sector as it is all about using certain artificial measures to hopefully make some fast money? Buffett has been a "value investor," not a fast-buck trader.
 
50% to 70% of stock trading is probably done using a quant strategy of some form

Well, interesting to imagine the market driven by quant strategies only. The input data for the trading software will actually be the outputs of the competitors' trading software (mediated by prices, vols etc). I think the victory will be with the pattern recognition techniques, behaviorial models.
 
Is there anything substantial to say about algo trading except that it has nothing to do with the inherent soundness of the company and/or the sector as it is all about using certain artificial measures to hopefully make some fast money? Buffett has been a "value investor," not a fast-buck trader.

I agree with Alain, I don't see anything of substance in the article either.
1. We don't have the metrics. Is it daily trading volume, is it yearly volume etc.
2. From algorithmic trading bucket, how much is pure automated trading, how much are trigger based transactions. To simplify, as you know, you can use a research part to trade long-term and use some thresholds to trigger automatic hedge of your portfolio. Or you can use wide volatility threshold that will generate automatically transactions. In both cases, you are trading just partially automatic.
3. How is input data generated? There is a large set of options, so different algorithmic traders can have different inputs, therefore they will not act synchronously.
4. Does algorithmic trading use balance sheet data? It can look at any indicator (e.g. PER) that will provide a component of company results. In the end, this is the bottom-line, first thing that any investor would analyze (including Buffet). So what is the problem?

There are many other questions, unless a formal analysis is presented, it cannot be convincing. Maybe it can work as beer talk ;)
 
4. Does algorithmic trading use balance sheet data? It can look at any indicator (e.g. PER) that will provide a component of company results. In the end, this is the bottom-line, first thing that any investor would analyze (including Buffet). So what is the problem?

The financial statements -- and the various ratios associated with them -- may look healthy but Buffett may shy away from the company or sector because of a long-term prognosis, some parts of which may not be quantifiable, which don't reveal themselves in financial statement analysis. Conversely, Buffett may invest for the long term in an area or company where the financial statements don't look promising because he's convinced the long-term prospects are excellent.

The point I'm making -- trying to make, rather -- is that focus on some or all of the indices you've listed seems to be part of playing a zero-sum numbers-oriented game that is indifferent to the underlying value of the company or sector. There may be "patterns" hidden in such analysis (or so some econophysicists insist), but the patterns have to do with outwitting other participants in a zero-sum game. This is the difference between calculated speculation and long-term investment. Again, I'm not saying anything startlingly original or "substantial." The upshot of market participation by such speculators has the effect of adding to market volatility. Again, passe.
 
So you see Buffet's approach as the only "fair/pure" investing?
In that case, I don't really see the discussion on this forum :)

There are many strategies that can yield a good profit. I don't see one being superior to another, especially since the risk of the strategy is hard to quantify. Therefore I don't see the market being flooded by variations of same algorithm.
Rather, I see the use of automatic trading in different conditions.

One of the points of my previous post was: how much of "algorithmic trading" is initial decision based? Basically how much of the trading involves taking a position in a stock that was not part of portfolio?
Assume following scenario. I research the market and I decide that I should invest in 5 mil in Microsoft, my target price is 20% , my range of accepted pain/profit is x etc.
Then I plug everything in an automated trading framework that will execute the trades without moving the market, it will rebalance, hedge position, will trade automatically if any condition applies etc. Is this algorithmic trading?

From my point of view it is not because it handles only one part, not the research ...
 
Therefore I don't see the market being flooded by variations of same algorithm.

I am curious to know the meaning of variations of same algorithm.

My take is that the risk of quant methods is that these methods cannot make any consideration of whether they can find the other side of the trade when things suddenly change. They are playing with their own assumptions. The smart players dumped their old strategies before other firms find the same ones and begin taking the same positions. But a lot of times it is just hard to tell when to get out of it.
 
I am curious to know the meaning of variations of same algorithm.

My take is that the risk of quant methods is that these methods cannot make any consideration of whether they can find the other side of the trade when things suddenly change. They are playing with their own assumptions. The smart players dumped their old strategies before other firms find the same ones and begin taking the same positions. But a lot of times it is just hard to tell when to get out of it.

"Model-Arbitrage" - the next big model?
 
Back
Top