Web Based Technical Analysis/Automated Trading Service

[Chaoskampf]

I'm in the process of developing a web based technical analysis service for crypto-currencies. It will have normal and automated trading capabilities in addition to advanced charting/analysis tools. As of right now, BTC-e and MtGox are the only exchanges that I've integrated into the platform. More exchange integration is sure to come, although I'm seriously considering dropping Gox support (I want to do my part to undermine its market centralization, however insignificant my impact might be). My bots are basically all written up and feature many high level functions, but I was just wondering what sorts of end-user analysis functionality people would want to see in a service like this. Other suggestions are more than welcome. Feel free to rip apart my idea.

[3Dfilament]

Some days before the bitcoin 2013 conference we saw Litecoin stabilize neatly around the $3 mark into a channel with a suppressed trading range, very narrow. And we saw this happen on the way up from the last correction to $2.5 levels where very large buy and sell walls were placed on the orderbook at btc-e which effectively obscured and made the bid/ask ratio metric unuseable unless filtered with a dynamic view. These large, moving buy and sell walls were articulated such that the trading ranges were constricted to channels and moved up to consolidate trading ranges sideways at the $3 levels.

One of the tools I find sorely lacking in trading Crypto is an interface to filter out the largest buy walls, or see only the buys/sells that are for example in an 8LTC to 45LTC range, and to examine other ranges of orders on the book might give one a perspective on a more accurate market depth, and market depths in different channels of the orderbook.

When we attempt to use the "short orderbook" on btc-e itself to gauge market depth, volume and movements, often we find the buy and sell walls heavily peppered with automated, high frequency, bot spam orders, or micro-orders. Filtering this out, and removing the largest buy/sell walls, would provide a much more accurate picture of buy/sell interests on the "long orderbook".

When performing statistical analysis, one of the first steps in finding a mean, or an average of course is to throw out the lows and the highs and average the middle values, and then of course there are many other metrics which can be derived from the orderbook. Of all of the tools in Crypto trading I find to be the least utilized it is the orderbook.

[Duetschpire]

Some days before the bitcoin 2013 conference we saw Litecoin stabilize neatly around the $3 mark into a channel with a suppressed trading range, very narrow. And we saw this happen on the way up from the last correction to $2.5 levels where very large buy and sell walls were placed on the orderbook at btc-e which effectively obscured and made the bid/ask ratio metric unuseable unless filtered with a dynamic view. These large, moving buy and sell walls were articulated such that the trading ranges were constricted to channels and moved up to consolidate trading ranges sideways at the $3 levels.

One of the tools I find sorely lacking in trading Crypto is an interface to filter out the largest buy walls, or see only the buys/sells that are for example in an 8LTC to 45LTC range, and to examine other ranges of orders on the book might give one a perspective on a more accurate market depth, and market depths in different channels of the orderbook.

When we attempt to use the "short orderbook" on btc-e itself to gauge market depth, volume and movements, often we find the buy and sell walls heavily peppered with automated, high frequency, bot spam orders, or micro-orders. Filtering this out, and removing the largest buy/sell walls, would provide a much more accurate picture of buy/sell interests on the "long orderbook".

When performing statistical analysis, one of the first steps in finding a mean, or an average of course is to throw out the lows and the highs and average the middle values, and then of course there are many other metrics which can be derived from the orderbook. Of all of the tools in Crypto trading I find to be the least utilized it is the orderbook.

That makes a perfect sense. When we see a 10,000 coin dump then pump on a $0.20 scale, the $0.01 move becomes very hard to monitor. I like the candle charts more as you can see which one is a sell and which one is a buy candle and how much difference is left out between sell/buy, but if we can get something similar to the charts on MCXnow and btc-e integrated together and make one candle colored orders but with the price separate to the volume, I think we'd have a good chart for the 24 hours period.

BUT, if we have the option to filter bots/micro orders and be able to scale the chart up to 7 days and down to 1 hour, we'd have a GREAT chart.

[Chaoskampf]

Some days before the bitcoin 2013 conference we saw Litecoin stabilize neatly around the $3 mark into a channel with a suppressed trading range, very narrow. And we saw this happen on the way up from the last correction to $2.5 levels where very large buy and sell walls were placed on the orderbook at btc-e which effectively obscured and made the bid/ask ratio metric unuseable unless filtered with a dynamic view. These large, moving buy and sell walls were articulated such that the trading ranges were constricted to channels and moved up to consolidate trading ranges sideways at the $3 levels.

One of the tools I find sorely lacking in trading Crypto is an interface to filter out the largest buy walls, or see only the buys/sells that are for example in an 8LTC to 45LTC range, and to examine other ranges of orders on the book might give one a perspective on a more accurate market depth, and market depths in different channels of the orderbook.

When we attempt to use the "short orderbook" on btc-e itself to gauge market depth, volume and movements, often we find the buy and sell walls heavily peppered with automated, high frequency, bot spam orders, or micro-orders. Filtering this out, and removing the largest buy/sell walls, would provide a much more accurate picture of buy/sell interests on the "long orderbook".

When performing statistical analysis, one of the first steps in finding a mean, or an average of course is to throw out the lows and the highs and average the middle values, and then of course there are many other metrics which can be derived from the orderbook. Of all of the tools in Crypto trading I find to be the least utilized it is the orderbook.

That makes a perfect sense. When we see a 10,000 coin dump then pump on a $0.20 scale, the $0.01 move becomes very hard to monitor. I like the candle charts more as you can see which one is a sell and which one is a buy candle and how much difference is left out between sell/buy, but if we can get something similar to the charts on MCXnow and btc-e integrated together and make one candle colored orders but with the price separate to the volume, I think we'd have a good chart for the 24 hours period.

BUT, if we have the option to filter bots/micro orders and be able to scale the chart up to 7 days and down to 1 hour, we'd have a GREAT chart.

You both make good points. While implementing this sort of functionality shouldn't be too difficult, figuring out what high/low volume orders are real and which ones are meant purely for manipulation might be a little tricky for the trading bot. Sometimes buy/sell walls are real. Even if they're not, they influence the direction of the market. While I could write a tool that filters down the order-books to a narrow spectrum of volumes, I don't know if I can incorporate it into the trading algorithm. My bots depend on these volume indicators to place their orders (and many different sub-functions within the trading bots use and process that information). If you're not interested in automated trading, then this isn't really an issue. You can just use that tool to filter the order-book and do your own analysis to place your orders by hand. Also, scaling the chart was one of the first features I implemented. I don't understand why that's not a standard feature on all the exchanges (it seems so basic and yet so incredibly useful). Does this address your suggestions or did I miss something?

[Loozik]

It will have normal and automated trading capabilities in addition to advanced charting/analysis tools [...] I was just wondering what sorts of end-user analysis functionality people would want to see in a service like this.

1. Ability to draw trendlines on a chart. No web service allows users to draw trendlines  

2. Multiple time frames, at least these ones: 1s, 10s, 20s, 30s, 1m, 2m, 3m, 5m, 10m, 15m, 20m, 30m, 45m, 1h, 2h, 3h, 4h, 6h, 8h, 12h, 18h, 1d, 1w, 1q

[Chaoskampf]

I don't want to publicize the entire functional capacity of the service until right before alpha testing, so that I don't provide competing developers (and there are definitely a wide variety of them, including corporate ones) with any undue inspiration. I'm sorry if this comes off as greedy or non-collaborative, but I've spent almost 4 months writing these algorithms and have been sacrificing my sleep and "free"-time to develop this. Although, I should describe some of the basic functions that are already incorporated into the program, and which any serious competitor will already be familiar with and use. This should give you guys a jumping off point for suggestions.

Current Analytical Functions:

1. Simple Moving Average
2. Exponential Moving Average
3. Moving Average Convergence-Divergence
4. Relative Strength Index
5. Support/Resistance
6. Aroon Oscillator
7. Stochastic Momentum Index
8. Chaikin Volatility
9. Bollinger Bands
10. On-Balance Volume
11. Linear Regression Trend Lines

There's plenty of on-chart drawing ability. As is, the charts only scale down to a 30 minute timeframe. I don't think I'll be making them any finer resolution than that. The upper bound for that scale is all-time data

[Chaoskampf]

also, follow on twitter for updates:

@Chaoskampf0

[Loozik]

Current Analytical Functions:

1. Simple Moving Average
2. Exponential Moving Average
3. Moving Average Convergence-Divergence
4. Relative Strength Index
5. Support/Resistance
6. Aroon Oscillator
7. Stochastic Momentum Index
8. Chaikin Volatility
9. Bollinger Bands
10. On-Balance Volume
11. Linear Regression Trend Lines

No Volume?

There's plenty of on-chart drawing ability.

Including trendlines?

As is, the charts only scale down to a 30 minute timeframe. I don't think I'll be making them any finer resolution than that.

It is a pity. Add at least 1 and 10 minute timeframe.

[Chaoskampf]

Current Analytical Functions:

1. Simple Moving Average
2. Exponential Moving Average
3. Moving Average Convergence-Divergence
4. Relative Strength Index
5. Support/Resistance
6. Aroon Oscillator
7. Stochastic Momentum Index
8. Chaikin Volatility
9. Bollinger Bands
10. On-Balance Volume
11. Linear Regression Trend Lines

No Volume?

There's plenty of on-chart drawing ability.

Including trendlines?

As is, the charts only scale down to a 30 minute timeframe. I don't think I'll be making them any finer resolution than that.

It is a pity. Add at least 1 and 10 minute timeframe.

You can draw whatever arbitrary lines you want. In addition, there are linear regression trend lines (as mentioned above) among other computed trend lines. There's obviously basic volume charting (I didn't think I needed to mention that), apart from advanced volume analysis like on-balance volume (as mentioned above). Any information that's available on BTC-e is already included in the program (last price, high, low, volume, order books)

[Chaoskampf]

If there's enough push for shorter timeframe scaling, I'll add 1 and 10 minute frames as well.

[Loozik]

If there's enough push for shorter timeframe scaling, I'll add 1 and 10 minute frames as well.

Good 

[whydifficult]

I don't want to publicize the entire functional capacity of the service until right before alpha testing, so that I don't provide competing developers (and there are definitely a wide variety of them, including corporate ones) with any undue inspiration. I'm sorry if this comes off as greedy or non-collaborative, but I've spent almost 4 months writing these algorithms and have been sacrificing my sleep and "free"-time to develop this.

It sounds like an awesome project. Should I take it it wil be a paid service (hence the 'competition')?

[Chaoskampf]

I don't want to publicize the entire functional capacity of the service until right before alpha testing, so that I don't provide competing developers (and there are definitely a wide variety of them, including corporate ones) with any undue inspiration. I'm sorry if this comes off as greedy or non-collaborative, but I've spent almost 4 months writing these algorithms and have been sacrificing my sleep and "free"-time to develop this.

It sounds like an awesome project. Should I take it it wil be a paid service (hence the 'competition')?

I'm not totally set on the payment plans, but yes. I think I'll only charge fees on the automated trading service, while the rest of the functionality will be provided free of charge (hopefully people will donate). I'm considering a flat rate on all automated trades that end in a positive gain. I'm thinking anywhere from 5-7% of that respective gain. No profits = no fees. You already have to pay trading fees on the exchanges, so I shouldn't be taking any more of your coins. Does this sound reasonable?

[3Dfilament]

3. Moving Average Convergence-Divergence

This, if only for litecoin, would be a huge leap forward in trading tools. Much like bitcoin.clarkmoody.

I visualize a slider GUI tool for those who like to place their trades by hand with a frequency of trades perhaps 1 every 1-5 days, not a daytrader.

To have some granularity, at any level, for filtering the orderbook would be, for all Crypto trading a giant leap forward. To create algorithms for automated trading based upon orderbook filtering might be a noble goal, but to gain feedback from the use of a broader, more basic orderbook filtering tool might provide data for building an orderbook filtering logic bot. If one can determine what the trader who executes on a longer timescale is looking for in the orderbook, then the development of algorithms to construct an automated bot which derives trading data from the orderbook might be more tangible for coding something more intense for automation.

[drwho88888]

If TA doesn't work for the major currencies, why do you think it will for crypto ?

And if you think TA works read 'Fooled by randomness' by Taleb

[whydifficult]

If TA doesn't work for the major currencies, why do you think it will for crypto ?

And if you think TA works read 'Fooled by randomness' by Taleb

TA tries to detect patterns in a market. I personally believe that markets for altcoins look different and are driven by different forces. This should be backtested though.

I've read the Black Swan by Taleb and its true that TA tries to predict what is going to happen while you can never know this for sure, however there are dangers everywhere: a lot of people on this forum have a lot of bitcoin (which is basically buy and hold, a dangerous strategy when crashes or even black swans can happen), a lot of people have little trust in fiat money (and argue the dangers also apply at the USD for example). So TA may be gambling with a lot of dangers, but it appears that in this crisis almost everything is.

I'm not totally set on the payment plans, but yes. I think I'll only charge fees on the automated trading service, while the rest of the functionality will be provided free of charge (hopefully people will donate). I'm considering a flat rate on all automated trades that end in a positive gain. I'm thinking anywhere from 5-7% of that respective gain. No profits = no fees. You already have to pay trading fees on the exchanges, so I shouldn't be taking any more of your coins. Does this sound reasonable?

I am just curious by the way. I think so, but it appears there is more and more competition for paid trading services and you have to differentiate somehow. Are you going to evaluate the profit after every buy-sell or on a timely basis?

[Chaoskampf]

If TA doesn't work for the major currencies, why do you think it will for crypto ?

And if you think TA works read 'Fooled by randomness' by Taleb

TA is entirely self-consistent and mathematically pure. In this sense, it works 100% of the time. They're just processed and re-processed numbers. The way you use those numbers is where I think you find the weakest link in efficacy. If by "work" you mean making a profit, then that's purely a function of what you do with this information. You could say the same thing about all charts, and all indicators. Why have charts in the first place? Why do we put volume bars underneath the charts? These processed numbers convey information. They don't have any personal prejudices. They are what they are, and deriving meaning from them is solely the responsibility of the humans looking at them. Here's the weakest link - the human. Some of them can derive more meaning from those numbers than others.

[whydifficult]

If TA doesn't work for the major currencies, why do you think it will for crypto ?

And if you think TA works read 'Fooled by randomness' by Taleb

TA is entirely self-consistent and mathematically pure. In this sense, it works 100% of the time. They're just processed and re-processed numbers. The way you use those numbers is where I think you find the weakest link in efficacy. If by "work" you mean making a profit, then that's purely a function of what you do with this information. You could say the same thing about all charts, and all indicators. Why have charts in the first place? Why do we put volume bars underneath the charts? These processed numbers convey information. They don't have any personal prejudices. They are what they are, and deriving meaning from them is solely the responsibility of the humans looking at them. Here's the weakest link - the human. Some of them can derive more meaning from those numbers than others.

Taleb explains in his book that TA is based around trying to predict the future based on history. Which can work a lot of times but you can never know it for 100% sure. He explains it by this Turkey who is fed everyday for a 1000 days. On the 1001th day the turkey is waiting for his meal as usual when something else happens. How could the turkey have known it was thanksgiving?



No matter how much TA the turkey would have thrown at the chart, he would not have been able to predict this. In this case thanksgiving is a black swan for the turkey. A black swan is something that you can't really predict and is never taken into account in the mathmetical model (else it isn't a black swan). This is more philosophical than it is practical, but Taleb says that if a black swan happens you can easily lose all your profits in a single event.

[Chaoskampf]

I'm not sure what you're arguing. Are you saying that it's better not to use TA because of the potentiality of an unforeseen black swan situation? Would not using TA put you in a better position during this situation? Would not using TA make the black swan go away/not happen in the first place? If that's what you're arguing, then I have to disagree.

[whydifficult]

I am not trying to bash TA at all, I am writing a TA bot myself to. But I just think you should be careful when you say:

TA is entirely self-consistent and mathematically pure. In this sense, it works 100% of the time.

The math is just an abstract model of an idea about how a market behaves. The idea may be pure but the market is not and unexpected things can happen that are not in the idea. Also market conditions change over time.

I do like TA and I use it myself, but you should never see it as a money making machine or something that will 'work 100% of the time' imo. It's just gambling backed by statistical data of the past, there is always risk.