Formalising common approaches to trading - page 29
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
You need to switch to a specific market, but it's probably too early :) But of course it's right - the microstructure of a particular market determines the possible ways to speculate. And in the subject of the branch the word "general"))) And most methods have a common basis. Although taking into account the specifics of the market should increase their effectiveness
P.S. But if you or someone else starts describing the specifics of a particular market, that would be good :)
It is worth discussing what lies behind specific tools and strategies (market logic). Especially since there is such a discussion and you can copy part of the discussion :)
What is the meaning of the moving average?
If to build an average of transactions at some time interval, taking into account the volume of transactions (by weighing), then such an average will be the level when the total profit of buyers at this interval will be equal to the total profit of sellers and equal to zero. Above this level, the buyers will be in the plus (and the sellers - with the same loss), and below - on the contrary. The classical average on discrete time intervals is just the arithmetic average price for the period, though it can approximately coincide in value with the above-described value and reflect the level of sellers' and buyers' profit at the segment, for which the average value was calculated. Let's assume, there are 2 transactions for some period (or all): the first for 100$10 lots and the second for 130$20 lots. The average price weighted by the volume of deals = (100 + 2*130)/3 = $120. At this price, profit of buyers = (120-100)*10 + (120-130)*20=0. The same for sellers. If the price will be above $120, then the buyers will be in the plus, while the sellers will be in the minus, and vice versa.
Apparently, the MA had such a meaning originally (the level of buyers and sellers interests balance), but due to the lack of Composers, they used an approximate formula, convenient for calculation. Then it became meaningful due to its wide use.
The volume-weighted average is called VWAP. At the same time, it is one of the most popular automated algorithms, used by institutions and provided by many brokers and platforms. Its purpose is to give good entry/exit prices for large volumes - i.e. to provide liquidity and minimize slippage and also to hide the intentions of speculators who want to profit from it. See for example how this method is advertised by Goldman Sachs http://premiacap.com/QWAFAFEW/morris_20020919.pdf
Then the average entry level of those who have entered by these algorithms will be very close to the volume-weighted average calculated for the frame of the day. As long as the price is above that level they are in the plus, when it breaks through from top to bottom they go to the loss. So they can fix profit/loss depending on the deviation of the price from this level (for example, percentage value or volatility). And there are a lot of tools using the same property of averages and fixing of losses/profits - CC envelopes, bollinger, etc.
By the way, all sorts of strategies based on Pivot Points are based on the same - the main pivot calculated for yesterday is the average entry level - the balance when buyers of sellers go from minus to plus and vice versa. And the target levels are expected levels of mass profit/loss taking. In this case they are tied to the volatility.
P.S. The approximate list of similar VWAP algorithms and companies that provide them
Goldman,Sachs & Co.
VWAP, VWAP Lite, TWAP, 4CAST (implementation shortfall), Percent of Volume, Piccolo (execution of small orders), Scaling, Gamma Hedge, Delta Hedge
BNY Brokerage
VWAP,TWAP
Citigroup Global Corporate and Investment Bank
VWAP, Cost Minimization, and Smart Market.
FlexTrade Systems Inc.
VWAP, TWAP, Target Volume, Sensitivity, Order Staging Model, Arrival Price
JPMorgan
VWAP, Target Volume, Advanced Pegging
Nomura
VWAP, TWAP, With Volume, Target Strike, Target Close, Wait and Pounce,
AutoReload, Spread (Pairs) Trader
with description
In general, the topic is a dead end. What is being attempted is to predict a storm in the cup. The main driver of any market movement is exogenous factors, in particular inter-market linkages and the herd effect. In other words, the price does not fall due to the lack of liquidity in the demand at the current level, and certainly not due to the location of limit or market orders but due to the external effect, the flow of liquidity from outside that in its turn affects the balance of forces in the specific market. In general, this is an attempt similar to predicting the market behavior by a technical indicator, though it is obvious that the market moves technical indicators, not vice versa.
Also, not a word has been said about market noise. That said, some understand noise to be an "unrecognised pattern", it is not. If we imagine that there is at least one trader (group of traders) on the market who trades deliberately or not randomly, we already have a certain amount of noise. The actions of these traders cannot be predicted. They cannot explain themselves why they made a certain deal. There is no regularity between their actions, so there is no tool that can predict this noise. Here I see it all in one pile. There is an attempt to analyze noise, which no matter how you analyze it, it will still be noise.
Here's a nice document by Merrill Lynch about the forex market. In the beginning of the document there is a certain analogy of division into exogenous and endogenous factors, but the emphasis is on exogenous factors(interest rates, trade balances, etc.).
P.S. already wrote, but still: I'm not suggesting to try to directly count and trade from the location of liquidity or market orders (this task is generally insoluble in most cases). It has been stated because an understanding of these things is necessary for almost any trading method, because everyone has to reckon with it. This is the elementary basis. The discussion is supposed to be mainly about speculative methods. You have this "herd effect"
P.S. has already written, but still: I am not suggesting to try to directly count and trade from the location of liquidity or market orders (this task is generally insoluble in most cases). It has been stated because an understanding of these things is necessary for almost any trading method, because everyone has to reckon with it. This is the elementary basis. The discussion is supposed to be mainly about speculative methods. You have a "herd effect".
For example, we should discuss the basis behind the levels - horizontal and inclined. And also about the channels they form. They have just that elementary basis in the form of liquidity zones formation.
For example, let's have an hour candle. While it is forming - all Limit orders from its low to high that were set earlier have been swallowed and some orders that were set less than an hour ago have remained. Of course, there are usually less of them than the orders above the high and low because they have been piling up at least an hour longer. And the longer the price has been hanging out in some range, approaching its edges and filling all Limit orders inside its range, the more Limit orders are piled up outside its boundaries forming horizontal support/resistance levels. Having learned this, some participants have started to place protective stops beyond these levels in the hope that the limit zones will prevent them from accidental triggering. As they continued to evolve))) some speculators realized that they could also place stops on the breakout of these levels. Because if the limit order zone is penetrated, the protective stops of the previous "generation" of speculators will start to work, and as they continue growing, stops of those who set the Limit orders at these levels will work too. In general, everyone sees these levels and tries to use the mass stereotypes in their favor, which evolve over time.
The classic definition of a trend in the form of rising local bottoms/decreasing tops has a similar logic: in an up-trend, buyers naturally concentrate near the local bottoms. This is reinforced by the stereotype of trading on pullbacks. The breakdown of the first lock/shortcut may trigger a series of pending stops as described for horizontal levels and the price will move higher to the next bottom, etc., causing a cascade of stops and significant downward movement.
Very interesting topic, thanks to the author. As I understand it the idea is to build a model (theory) based on some first principles, in the hope that this theory will not only explain the known phenomena but also predict new ones, not yet noticed by the general public. It is reasonable to expect that for new phenomena the probability of different scenarios may differ significantly from 0.5/0.5. Although ... how reasonable is that? However, it is quite a philosophical question.
At any rate, the public interested in this question is so large and advanced that the probability of detecting and/or somehow prolonging the existence of "private" phenomena seems very slim.
As for already known phenomena, it is possible to build more simple, phenomenological models for them, i.e. models that use certain properties of the market without going into mechanisms of their appearance. In fact, this is what the vast majority of people trying to play the market do.
I do not mean that the task is not interesting, on the contrary. But a qualitative rather than quantitative prediction of a phenomenon may be a more realistic result (albeit quite fantastic). And for its verification and (with luck) use the phenomenological model will be practical.
As for the already known phenomena, it is possible to build simpler, phenomenological models for them, that is, models that use certain properties of the market without going into the mechanisms of their emergence. In fact, this is what the vast majority of people trying to play the market do.
I do not mean that the task is not interesting, on the contrary. But a qualitative rather than quantitative prediction of a phenomenon may give a more realistic result. And for its verification and (with luck) use the phenomenological model will be practical.
You can go into details with different "depths". For example, much less is known about the currency market (than about exchange markets), in order to estimate precisely the methods of the currently traded participants, although there is information and common grounds too. What this market logic is needed for is to understand where and what to look for and how to build "phenomenological models", eventually trading systems, how to test them, what to optimise, etc. And not to just go through the indicators and their parameters randomly combining them, which is the way to fit in. And without details any statistical research gives out "the average temperature in the hospital". Justification of market logic is an additional confirmation of the robustness of the pattern found and often reduces the time it takes to find it.
I.e. there are two directions - either digging in the direction of market logic, or purely statistical. imha, both of these methods can be used in a reasonable combination. But still the basis of all speculation is to prevent other people's mass trades, which implies understanding their logic - what they are based on, what unites them, etc. Otherwise the search for a robust system is a search for something in a haystack :) And of course, statistics/testing. Without it you can go far in your fantasies. Including in the construction of phenomenological models. Practice confirms theory, but theory is also needed within reasonable limits))))
By the way, not all TA methods use endogenous factors, some use external factors. For example, classical trend following. It is based on trying to catch significant price movements caused by factors that are not yet in prices - new money coming into the asset. The arrival of money is caused by external factors - macroeconomic or for example the arrival of a new strategic investor. Trend-following is an attempt to catch such moments without going into details and reasons. It is counting on the fact that these processes are stretched out in time, not discrete. This allows us to hope that the trend will continue rather than end. Also, the trend development can be strengthened by exogenous factors in the form of speculators, who will trade seeing this movement - such as traders, impulse traders, etc. In general, the reasons are not important - we are interested in the tendency of the market to be trendy. Pastukhov's H-volatility dissertation and Shiryaev's reports on this subject provide a good mathematical basis for trend following. Their research is based on conventional statistics based on delta modulation of price series by price. These are tools such as Renko, Kagi, P&F, and ZigZag. If the market is trending, the average swing length should be more than 2*H (modulation parameter: Renko size or ZZ parmeter). If less, then markets tend to return (or more simply to form a flat). For a random walk this value =2H. Here is their work and speeches on the subject. This is a universal formalization of methods based on delta modulation of price.
But in modern markets for most instruments this H-volatility is exactly equal to 2H. Except if we consider separately up-swings for developed countries stock markets - they tend to trend, which is explained by global growth of these economies and inflation. I.e. pure trend-following does not work now, or its profitability is no better than a buy&hold strategy. Of course, we can try to identify situations when the market is trending or flat (filtering), but it is not a classical trend following and there is identification of specific market processes behind the filter. Or, if they are versatile enough and this name is more convenient, we can call it phases of the market :)
Rules such as "cut losses and let profits grow" also belong to this type. For trend markets this rule itself has a satistical advantage. It also makes sense to add to the profitable one as the trend is confirmed. In this regard interesting book by A. Simpson "The stock market ghost" where he focuses on rules of phased entry and exit - a gradual exit from weak positions - the target is not reached, the next extremum is not broken through within a certain time frame, the reinforcement of winners (shares to profitable positions), etc. Discussion
The book itself is here In principle the book is interesting and in fact these rules for effective trading in trend markets.
For trend markets, this rule in itself carries a satistical advantage.
The price movement has a wave structure, a rush of demand replaces supply and back again and it's all in a cycle. Step by step, waves of different sizes, levels are formed out of this chattering. The higher levels are formed from the lower ones according to the same rules as the lower ones. We obtain the fractality consisting of N levels (e.g. timeframes). The more these levels N, the greater the chances that the trend will be in the future (If the previous history was formed, say, 8 levels, it is unlikely in the future with unchanged market (well, if the market economy does not become planned or other) the number of levels down to 1, ie unlikely the price will be in the channel, with an equal step). At the same time, there is a model when one can lose if the flat will be longer than impulses, i.e. if one cannot manage to compensate losses with an impulse, the system will lose (for a trend system). Curry trade, here may help a little in the choice of instrument.
On the topic, we need a chart or a table, we need a tree! Gradually filling in/correcting by common efforts.
We can start like this
Maybe replace internal and external factors with TA and FA? Or define the composition of internal and external factors, because it is not clear who means what based on our activities.