[permanentjaun]

Astro,

You made a good point just now which is something similar to what I've been thinking about recently. What are we trying to do with MA's? We're looking for trends and momentum. Thats why they don't work very well in ranging markets. Their lag prohibits them from working in ranging markets or signaling an early trend if it happens too quickly.

1. They need a new candle formation before they can be used. It may be safer to play a longer chart time frame, but you leave yourself open to a lot of possible pip movement. Lag.

2. They only take into account 1 price; either the open, high, low, or close. They disregard all other information. Is that a flaw or a strength? I think it may be a flaw. For instance, what happens when a price drops tremendously in one candle, say 50 pips. If you're playing an MA cross system the cross may not happen till the next candle or two. That lag isn't so bad. What's bad about it is if the next candle or two are actually positive movements. The MA still crosses down because of the lower prices, yet the current movement is very positive. You enter short even though the price is showing good up movement. Lag and whipsaw.

Therefore I think we need an indicator of a different sort. It will be an oscillator type. Looking at the traditional oscillators I don't think they'd work very well. They have the same problems of MA's as they are based solely on the open, high, low, or close and not all of their ticks. They are also oscillators based off of price and MA divergences.

What do you all think of a simple oscillator based on a proportion of upticks vs. downticks? It must be based on ticks solely. If its based on OHLC then it must wait for new candles to form. I want it to be a dynamic time series indicator. Let me explain:

Say you're looking at a 10 minute chart. The price has been trading flat for 10 candles. Each candle has had a total of 10 ticks in them for their trading. (I realize this isn't very realistic but I'm trying to keep the math simple.) Because the trading has been flat we can say that out of the 100 ticks total, the proportion of up and down ticks has been 1:1. Now all of a sudden in a new candle the price begins to trade down. It drops 30 ticks. Instead of using all the original 10 candles ticks it only uses 100 ticks total. So this new candle, which uses 30 of the 100 ticks, knocks out 3 of the first original 10 candles. Its only using the latter 7 candles of the 10 and the 30 ticks from the new candle to come up with a value for the indicator. If in 10 minutes it moved 100 ticks then only the information in the last 10 minutes would be needed as the inactivity of the previous candles is now obsolete in both the formula and actual usefulness since there is a new volatility in the currency.

The proportion doesn't care about time, only pip movement. If you entered a play and it went up 1000 pips in 5 minutes would you still hold on because youre a position trader and think it still has room to go based on the charts? I'd care about the pip movements and get out with my profit. If it went up that fast in 5 minutes then it may do the same going down, in which case all indicators are thrown out the window for the volitility being too much for their formulas.

Sorry for explaining it in such a complicated way. Basically I want to know the proportion of up vs down ticks over the last X number of ticks. I don't want to care about time frames.

Why do I think this would be useful?

1. Lag would be reduced. It would use only the information useful to the current situation. Go back to the situation I presented earlier where the price dropped to almost form a MA cross but rallied from the bottom even though the later cross still signaled a short sell. The MA was only concerned with one price, OHLC, and didn't compute the severity of the move.

I believe the simple proportion indicator would have either enabled you to get into the drop sooner as previous ticks before it were replaced with downticks, or been able to save you from entering a short position as it rallied from the bottom since previous downticks in the drop were now getting replaced by upticks in the rally.

2. It would easily allow a computer to recognize when a market is ranging and for it to stay out of a position. Even more importantly is it would allow a computer to recognize a breakout. It wouldn't need to know at what price level a currency has been bouncing off several times. It would just recognize that the breaking of that support/resistance is bringing in new up or downticks which would sway the value of the indicator.

Right now it's just a simple oscillator. Perhaps just (# of upticks +1)/(# of downticks +1) = proportion oscillator. You need the +1, at least on the bottom, just so that it doesn't go to undefined when there are no downticks in the range specified. (Improbable but not impossible depending on how many ticks you want in the range) Or the oscillator could be two points with a midpoint line. If you want a range of 100 ticks. The midpoint is 50. As more upticks begin to take place it goes to 60 and downticks drops to 40. Make a histogram in between the two points which changes color depending on if the bulls or bears have control. Although, I think it would be harder to program an EA on this method. Not impossible though.

I don't see any current oscillators doing this though. None other takes into consideration all information and not just OHLC or MA divergences. Also, I don't understand why oscillators have to be limited to a certain range. Why does the RSI have to stay inbetween 0 and 100? What is the purpose of formulating it to do so? I would believe with a simple proportion indicator that it would auto correct itself to stay in a range.

For sake of argument say you're using a 1000 tick range. It goes up 1000 ticks with no downticks. (highly improbable) The value would be 1001/1 using the first method mentioned. Value = 1001. It begins to turn and has just 1 downtick. Value = 1000/2. Value would correct to 500. Another downtick to 999/3 = 333. It corrects itself quickly as more downticks begin to signal a reverse in trend. With a 1000 range the limit value for a down trend would be 1/1001 which approaches a limit of 0. This could be reversed easily so that downtick proportions could reach more managable lows and numbers when making decisions rather than seeing .0009099000 on your computer screen.

There are a couple problems with this oscillator however. For one, if you like looking at a 4 hour chart the tick range would probably need to be in the tends of thousands if not hundreds of thousands. We all know computers can easily compute this, but do our brokers do it? Do they offer a server such that I could use a large number of ticks from 1000 hours ago? Do they bother to keep track of an uptick 3 months ago? The information is simple and could be stored easily, but would they do it? Perhaps their computers automatically already do it. Stock chart sites offer historical quotes and charts such that you can see a 5 minute chart on a certain date months if not years ago. Is forex the same way?

Another problem is the actual formula itself. Consider the scenerio of a flat trading currency. It then begins to fall. This indicator introduces a new lag where it wouldn't give a signal if the new pips are the same as the pips it replaces. Meaning if the newest, or 100th tick in the series is down and the oldest tick, or 1st in the series, is also a downtick. There would still remain the same proportion of upticks vs. downticks in the equation. It would lag in that sense, but we must assume that in time upticks would get removed if a true downtrend were commencing.

Another problem is what if its not an up or a downtick? What if it downticked because of a sell at the bid and then someone else sold at the bid before the bid/ask prices could change? It shows another tick, but sideways. Do we disregard it and only concern ourselves with up and downticks? Do we credit it with the last type of previous tick, in this instance a downtick? Some things to consider.

So I ask you all to tell me if this would be a useful indicator or not. If it might seem useful, are there any oscillators similar to it? If not, should the oscillator stay that simple? Are there any ways to improve it through the formula? I feel stupid asking for such a simple indicator. I've only just started to think about such an indicator today however.

I base all of this on my theory that indicators could be more useful if they used more information rather than just OHLC or some form of it. For example, a price moves from 10 to 20. Would you feel more comfortable entering a long position knowing that it easily ran to 20 with little retrace or that it went to 13, back to 11, up to 15, down to 11, up to 14, down to 12, up to 17, down to 14, up to 17, down to 13, up to 19, down to 17, finally up to 20. I believe this would help us determine how much of a battle is really there. If it ran from 10 - 20 easily in one candle then I imagine the next candle would have a good chance at being another large bodied white candle. If it struggled to get from 10 to 20, like in the second example, then I assume it has a higher chance of forming a spinning top, hanging man, shooting star, gravestone doji, or any other candle I've forgotten. WHO IS WINNING THE BATTLE? Answer that question and join their side. That's all we really need to know to make a successful trade yea? Look forward to hearing comments. Matt