Sorry to post this in two places (first my journal, now here), but it is relevant. Following on from bapxyz's post:
The latest edition of E-Forex magazine has a couple of interesting articles. The first (pages 35 to 38) looks at the issue of one-click trading, global liquidity and slippage. It's not really relevant to global macro/position trading, but its pretty cool. Here are some of the interesting points:
- Ideal case spreads: Simulating the 'best of the best', the authors put together the best prices from four FX marketplaces (Dukascopy, Lava, Hotspot, and Currenex) and a selection of leading bank liquidity providers, to give the prices of an ideal marketplace.
The experiment ran on 18/10/06 from 14:10 GMT to 15:10 GMT, and the results are pretty cool: The EUR/USD spread is just 0.903 pips, GBP/USD is just 1.52, USD/JPY 1.05 and USD/CAD 2.20. In terms of spread distribution, EUR/USD was at 0.5 pips for 21% of the time, 1.0 pip for 65%, 1.5 pips for 10%, and 2 pips for just 4% of the time. The average bid life span for EUR/USD was 3.01 seconds (ie before it the price changed).
I guess this experiment didn't take place over a key data release, because the upward spike in the spread would have skewed the results. I think it would have been interesting if it was carried out for a full trading session.
- Trader delays: Given the speed with with FX market prices update, the ability of the human trader to react to these changes needs to be considered. There are two delays in the trader's response: First, there is the internet connection delay between you and your broker, and then there is the actual ability of the trader to recognise and respond to the data.
The physical location of the trader relative to the broker is less important than the trader's response time, but it still plays a role in adding to the response lag. In a table of ping response times between the Duksascopy server in Switzerland and in other countries, the delays were: UK (0.030 seconds), USA (0.095), Canada (0.120), Brazil (0.280), Japan (0.310), China (0.450).
Dukascopy also have a great tool that tests traders reaction times. The article says the average reaction time of a human trader, in 'one-click' activity, is 0.4 seconds. Have a go. My average time on the second go was 0.37 seconds, with a single fastest response time of 0.26 seconds.
They offer the following exercise:
- Test your own reaction delay.
- Measure your ping connection to your broker
- Add them together to give your total delay time.
Why is this important? Well, with FX prices changing so fast, and given these additional lags, we are faced with the possibility of slippage - ie that is clicking on a price that is no longer valid and getting filled at a different price. The article says that with a total delay of 0.4 seconds, the trader has an 8% chance of slippage, but this is reduced to 4% if the trader uses a local broker and has a delay of just 0.35 seconds (not much difference, eh!).
They recommend using a broker with slippage control on one-click trading, or using the 'place bid/offer' function in marketplaces, to ensure you get your always price.
The authors conclude with the view that spreads will continue to tighten, with faster and faster price updates, that more tools will be made available to avoid slippage, and that the FX industry will compete to be as geographically close to the trader as possible.
In the meanwhile, if you are a rapid-fire trader and have a system where every millisecond counts, I recommend serious traders practice your cat-like reflexes daily (a good excuse to keep your playstation or X-Box!) and move home to be as close to your broker as possible. [img]images/smilies/smile.gif[/img]
http://www.currenex.com/images/eForeX-logo.gif
The latest edition of E-Forex magazine has a couple of interesting articles. The first (pages 35 to 38) looks at the issue of one-click trading, global liquidity and slippage. It's not really relevant to global macro/position trading, but its pretty cool. Here are some of the interesting points:
- Ideal case spreads: Simulating the 'best of the best', the authors put together the best prices from four FX marketplaces (Dukascopy, Lava, Hotspot, and Currenex) and a selection of leading bank liquidity providers, to give the prices of an ideal marketplace.
The experiment ran on 18/10/06 from 14:10 GMT to 15:10 GMT, and the results are pretty cool: The EUR/USD spread is just 0.903 pips, GBP/USD is just 1.52, USD/JPY 1.05 and USD/CAD 2.20. In terms of spread distribution, EUR/USD was at 0.5 pips for 21% of the time, 1.0 pip for 65%, 1.5 pips for 10%, and 2 pips for just 4% of the time. The average bid life span for EUR/USD was 3.01 seconds (ie before it the price changed).
I guess this experiment didn't take place over a key data release, because the upward spike in the spread would have skewed the results. I think it would have been interesting if it was carried out for a full trading session.
- Trader delays: Given the speed with with FX market prices update, the ability of the human trader to react to these changes needs to be considered. There are two delays in the trader's response: First, there is the internet connection delay between you and your broker, and then there is the actual ability of the trader to recognise and respond to the data.
The physical location of the trader relative to the broker is less important than the trader's response time, but it still plays a role in adding to the response lag. In a table of ping response times between the Duksascopy server in Switzerland and in other countries, the delays were: UK (0.030 seconds), USA (0.095), Canada (0.120), Brazil (0.280), Japan (0.310), China (0.450).
Dukascopy also have a great tool that tests traders reaction times. The article says the average reaction time of a human trader, in 'one-click' activity, is 0.4 seconds. Have a go. My average time on the second go was 0.37 seconds, with a single fastest response time of 0.26 seconds.
They offer the following exercise:
- Test your own reaction delay.
- Measure your ping connection to your broker
- Add them together to give your total delay time.
Why is this important? Well, with FX prices changing so fast, and given these additional lags, we are faced with the possibility of slippage - ie that is clicking on a price that is no longer valid and getting filled at a different price. The article says that with a total delay of 0.4 seconds, the trader has an 8% chance of slippage, but this is reduced to 4% if the trader uses a local broker and has a delay of just 0.35 seconds (not much difference, eh!).
They recommend using a broker with slippage control on one-click trading, or using the 'place bid/offer' function in marketplaces, to ensure you get your always price.
The authors conclude with the view that spreads will continue to tighten, with faster and faster price updates, that more tools will be made available to avoid slippage, and that the FX industry will compete to be as geographically close to the trader as possible.
In the meanwhile, if you are a rapid-fire trader and have a system where every millisecond counts, I recommend serious traders practice your cat-like reflexes daily (a good excuse to keep your playstation or X-Box!) and move home to be as close to your broker as possible. [img]images/smilies/smile.gif[/img]
"Always bet on black"