Deadline June

[DashRiprock]

Just some thoughts, with no intention to get into an argument.

Adamus and I have been trading since the 1990s (discretionary and automated) so it's not like we haven't observed the markets, because it would be impossible not to peak at them even by mistake.

Another thought: are we sure that looking at the live markets is much different than looking at them once that action has become recorded history? I think you can learn about how the markets move even if you look at their recorded history, rather than just live (maybe even more). Just a thought, because you might have meant something different.

me too i dont want to start fights especially not on someones elses blog.

sorry but there are some funny things here

10 years of trading but still no beliefs about markets?
you can trust your mind to make the right decisions (or any decision) all the time?
you can learn as much about markets and trading from looking in the past instead of being in the present?

i can only speak from my life in trading but that is very strange to me :|

 

[Adamus]

Dash, you are too slapdash by far. You certainly like fighting, although it wasn't you who started it. I think we've already put this "beliefs about markets" part of the argument to bed, so either say something new or move along.

 

[DashRiprock]

Dash, you are too slapdash by far. You certainly like fighting, although it wasn't you who started it. I think we've already put this "beliefs about markets" part of the argument to bed, so either say something new or move along.

i definitley didnt try to start to want to fight back at anyone and specially not on anyones personal blog.

sorry if you think that disagreeing with you is fighting. i wont post anymore on your blog, i still think my post about starting with a piece of paper and what you want from a trading strategy was a good one like i said in our emails.

but bottom line is that if you have been trading for 10 years or more and you still dont have beliefs or even understand the importnace of beliefs or have ideas about what you want to get from your strategies then basically you are pissing in the wind.

 

[Yamato]

As usual here's my thoughts, with no intention to get into a heated argument.

Regarding this:

you can learn as much about markets and trading from looking in the past instead of being in the present?

I am sure you agree with me but there's some misunderstanding about semantics and what we are saying. If you are a mechanical trader and back-test systems, of course you will agree that the past teaches us as much if not more than looking at live markets. First, we can only look at so many markets for so many hours. Second, and most important, the "present" becomes history in one second. As soon as today is over, that is called "history". So basically the only difference between live markets and historical databases is the element of emotions, which make us react to live markets in a different way. However, since we are automated (or call it "mechanical" as some do) traders, emotions are out of the picture, so i believe you could even totally skip live trading and, as an automated trader, you would be ok and could make money. I even know a person who has never placed a single discretionary trade and makes money with his automated systems.

 

[brettus]

As usual here's my thoughts, with no intention to get into a heated argument.

Regarding this:

I am sure you agree with me but there's some misunderstanding about semantics and what we are saying. If you are a mechanical trader and back-test systems, of course you will agree that the past teaches us as much if not more than looking at live markets. First, we can only look at so many markets for so many hours. Second, and most important, the "present" becomes history in one second. As soon as today is over, that is called "history". So basically the only difference between live markets and historical databases is the element of emotions, which make us react to live markets in a different way. However, since we are automated (or call it "mechanical" as some do) traders, emotions are out of the picture, so i believe you could even totally skip live trading and, as an automated trader, you would be ok and could make money. I even know a person who has never placed a single discretionary trade and makes money with his automated systems.

I think it's impossible to make an automated system without watching it live, but i believe both Adam and Travis do watch it. I do think looking at times where theres economic data and market opens and closes, could be an interesting addition to their arsenal.

 

[Yamato]

Yes, I did get about half of my trading ideas while looking at live markets. But roughly another half of my ideas I got by testing things on historical databases. And, as I said, I know a successful automated trader who never placed a discretionary trade, and so he probably did not spend much time looking at live markets. Thanks for the tip ("looking at times where theres economic data and market opens and closes").

 

[Adamus]

automated trading and news events

Yes, I did get about half of my trading ideas while looking at live markets. But roughly another half of my ideas I got by testing things on historical databases. And, as I said, I know a successful automated trader who never placed a discretionary trade, and so he probably did not spend much time looking at live markets. Thanks for the tip ("looking at times where theres economic data and market opens and closes").

Running your automated trading (or not) is a much debated issue with several pros and cons on either side of the argument.

Obviously if I spent the time to build up experience on the market reaction to news, I could probably improve the performance of my automated systems. However the backtesting I do involves simulating trading through all news events, so I do that with my live systems too. That's my opinion.

Secondly, one hears stories about traders automating systems just to trade news events by using some kind of news headline scanned input into their system, to get a superfast reaction to the news. So it seems the news headlines are becoming just another indicator. If that's the case and I could code it, then if it helped, I'd use it. If it didn't help, just like a triple MA cross-over signal might not help, then I'd not use it.

Of course the last alternative would be to trust my systems to a discretionary trader who knows how to trade the news. He could have a big red button on his desk and if he sees the news would be bad for the system, he hits the button and the system shuts down. I wouldn't trust myself to do that myself.

 

[brettus]

Re: automated trading and news events

Running your automated trading (or not) is a much debated issue with several pros and cons on either side of the argument.

Obviously if I spent the time to build up experience on the market reaction to news, I could probably improve the performance of my automated systems. However the backtesting I do involves simulating trading through all news events, so I do that with my live systems too. That's my opinion.

Secondly, one hears stories about traders automating systems just to trade news events by using some kind of news headline scanned input into their system, to get a superfast reaction to the news. So it seems the news headlines are becoming just another indicator. If that's the case and I could code it, then if it helped, I'd use it. If it didn't help, just like a triple MA cross-over signal might not help, then I'd not use it.

Of course the last alternative would be to trust my systems to a discretionary trader who knows how to trade the news. He could have a big red button on his desk and if he sees the news would be bad for the system, he hits the button and the system shuts down. I wouldn't trust myself to do that myself.

Alot of people make money fading certain news events. If you could somehow code that. Mainly the medium and low importance news would be best to fade. I wouldn't fade GDP, interest rate decisions etc, but i'm sure you could figure which news works best.

 

[Adamus]

News and automated trading, more

I can't imagine it would be possible unless you paid good money for access to a decent newsfeed and one with history on file to back it up. Plus it's not something I could just plug into NinjaTrader. Well it could be if you cared to program something to work with NinjaTrader but I wouldn't want to commit myself to NinjaTrader that deeply.

 

[Adamus]

Sunday afternoon planning session

After reading "The Seven Habits of Highly Effective People" I decided that I too should become highly effective - and part of this involves prioritising one's tasks for the week ahead using the Seven Habits system and calendar / diary. However I haven't got around to buying the calendar /diary yet - I see a bit of a vicious circle building here.

Since one of my major roles in life is trading (you define all your roles using this system), I have to prioritise my long term goals, and then reduce that to medium term (i.e. what I want to achieve by the end of the year) and then short term - this week.

So, long term goals include:

(1) Work out a specific figure that I want to make from trading at which point I will walk away from it and do something else
(2) Using my software development skills, build a trading platform that does all the good things I need that I know from the various different trading platforms I've used or evaluated, e.g. TradeStation, NinjaTrader, TradingBlox. Obviously way too big a job for one person, I'd like it to be an open-source collaborative project made up of trader-developers.

Medium term goals:

(1) Find or invent, write, test and implement trading systems firstly to cover monthly outgoings and secondly to increase the capital in my trading account. At this time this is a question of keeping my living costs as low as possible, and secondly putting as many systems as I can onto NinjaTrader. Since I have no idea where the limit for NinjaTrader lies, this is pretty much open ended. I can also set up a second computer anyway and buy another NinjaTrader license. The main bottle-neck on this is the rate at which I can produce systems.

(2) Software - current state: I have an application which can import trades/PnL originally for TradeStation but modified for any trade history, and run portfolio analysis against it. It also takes a TradeStation format order batch file and transmits them to IB TWS. I think my medium term goal should be to get this up and running without the database performance issue it currently has dealing with millions of trades.

Short term goals:

(1) Gap Filler system: run it tonight using either the 1 tick bars that allow me to buy or sell practically on the second/third tick after the IB open (so they say - probably the 10th tick knowing my internet connection) or run it using 1 mins bars that I use for doing the backtesting.

(2) Other low hanging fruit in the systems ideas list I have: Joel Rensink's One Night Stand and First Strike systems.

(3) Get some kind of definitive response from British Telecom on the network problems causing the disconnects in the IB TWS logs (that were responsible for missing $800 of profit last week)

(4) Software (must find a name for it): set up the project testing batch run and code a database performance test which fails if it falls below a desired level. That might take a bit of research.

(5) Sofware 2: get Apache and Tomcat running together on VPS

 

[Adamus]

Gap Filler system - added filters, done risk optimisation

There were several fundamental decisions to make about how to trade the gaps created by weekend price movement from Friday's close.

Most of the ideas stem from the traders writing about how they trade the gap on the S&P index futures. I didn't find any material from people trading forex weekend gaps. The S&P traders trade every day's gap, not just over the weekend, using the official S&P market's opening and closing times despite the fact that nowadays the market isn't even closed in-between during the week because Globex is running.

The first decision was whether to base the stop size on Friday's range or on the average range e.g. 21 bar daily ATR. The question was logically does Friday's range have more weight re the stop placement than the average of the last 21 days?

I decided to see what was more reliable in the backtest performance.

Secondly, I wanted to filter out the small gaps. There's no point in taking on a big stop risk for the sake of a tiny reward that doesn't even cover the commission and slippage. I decided to set the filter not to take gaps that were less than 5 points in size.

Thirdly, there are theories that big gaps are dangerous to trade because they have a lower probability of filling, although they obviously bring in more points when they do. Brett Steenbarger - a relatively well known TA pundit - said on his blog that gaps of 40% of the previous day's range on the S&P is the magic number. Any higher than 40% and less than 50% of them filled.

I optimised the stop size as a % of the previous days range and the gap % size limit for the different forex pairs. I got much better looking optimisation result surfaces compared to the previous system I tried optimising. Jpgs attached. This gives me more confidence in the optimisation process and general robustness.

There was a fourth very TA oriented approach to filter the trades, which I haven't done and probably won't, described in the Futures Mag where they defined 5 different 'zones' that the open fell into - all based around the OHLC of the previous day, and all with historical probability of filling when occurring. Too technical for me.

Another filter I haven't tried is an MA to filter out trying to fill gaps against a strong trend. I guess I mean a trend strength indicator too.

 

[Adamus]

Re: Gap Filler system - added filters, done risk optimisation

The first decision was whether to base the stop size on Friday's range or on the average range e.g. 21 bar daily ATR. The question was logically does Friday's range have more weight re the stop placement than the average of the last 21 days?

I decided to see what was more reliable in the backtest performance.

I could see no difference in performance between the 21-day ATR-based stop and the stop size based just on Friday's range. Of course these stats are slightly misleading - they are the average or the sum of the constituent instruments, not the proper portfolio stats.

Instrument			Friday range	ATR(21)
Profit Factor			1.69			1.66
Average Trade			57.97		56.33
Total Net Profit			196355		190785
Profit per Month		169			165
Max. Drawdown			-5053		-5031
Profit/DD				3.84			4.25
Total # of Trades		3387			3387
Percent Profitable		86.6%		87.0%
Ratio avg. Win / avg. Loss	0.26			0.25
Average Winning Trade	164			163
Average Losing Trade	-623			-658
Max. conseq. Losers		2			2
Largest Winning Trade	4660			4660
Largest Losing Trade		-4160		-3275

Since it seems more likely to me that the previous 21 days volatility is more likely to influence the next days volatility compared to the previous day alone, I'm going to go with the 21-day ATR.

 

[Adamus]

TurningPoints Improvements Needed

I'm going to start saving charts from NinjaTrader with the Chart Trader showing the TurningPoints strategy running in mid-trade that should never have happened.

Hopefully at some point in the future I'll find ways to program the strategy better.

This trade attached is probably going to be stopped out soon. It's a disaster from the start.

The enter-trade came way too late. The 'Turning Point' name refers to the strategy of spotting turns in the market, and the one on this market was the low of the day yesterday (US style dates on x axis).

So it turned, made a new low but the simple MA cross-over that confirms the turn didn't happen until 2 big bars had gone past.

So maybe big bars like that should be ignored.

Secondly, the entry was too high already because (a) it's entering against the trend down from the top on 27th, and (b) there's a pretty clear down channel too which the entry point is at the top of.

Thirdly the stop is way too close. I have no idea without looking at the code why the system would have chosen to put it there and not at the low yesterday.

Snafu.

 

[Adamus]

Re: Gap Filler system - not satisfied with simple system, looking at improvements

Just to recap, the system to trade the gaps after the weekend from Friday's close is structurally 99% complete now.

I had to program all the forex pairs I plan on trading into one NinjaTrader7 strategy as 'additional time series'. This gives me stats for the whole set of P+L as a portfolio.

I have to decide which switches and filters to apply. There are several to choose from as I have already outlined.

(1) how much of the gap to aim for when placing target order - I haven't varied this and I don't think it's that important, but I chose 90% on the whim that I don't want to see that the price reached the target but the order was never filled.

(2) filter out the small gaps. I won't take gaps that were less than 5 points in size.

(3) filter out the huge gaps. I optimised this - 40%, 50%, 60% etc of Friday's range. It didn't make much difference, I figure I can fix this at some level or just drop the filter.

(4) whether to base the stop size on Friday's range or on the average range e.g. 21 bar daily ATR - the results were no different, decided on using he 21-bar daily ATR.

(5) what % of (4) should be used for the stop size. I optimised this too, using a % of the previous days range. I think this is the most important parameter.

(6) Friday's range can be divided into 5 zones with different probabilities that the gap will fill. If Friday's candle is 3 zones, (candlestick = 1 zone, wicks = 1 zone above & 1 zone below candlestick) the 4th zone is above the high and the 5th below the low. I don't know if the technical relevance of this in the S&P market is equally valid in forex though. I doubt it somehow.

(7) use an MA to filter out trades against the trend, maybe just a strong trend - in which case I need a trend strength indicator too

(8) another idea from this S&P gap trader is to place the entry trade as a stop away from the Open, to benefit from any market move away from the gap initially. This requires optimisation, or perhaps just placing it half way from the Open to the stop loss.

Here are the optimised results for the basic system - with 1,2,3,4 & 5. The results include a proper drawdown figure for the account, as opposed to the dumb NinjaTrader average drawdown. It was a pain in the posterior to get it working in NinjaTrader, but now I have it frees me up from loading it into Excel every time I want a serious statistic for my results.

Performance		All Trades
Total Net Profit		$152685.00
Gross Profit		$409885.00
Gross Loss			$-257200.00
Commission		$0.00
Profit Factor		1.59
Cumulative Profit	$152685.00
Max. Drawdown		$-22555.00
Sharpe Ratio		0.42
		
Start Date		01/12/2000
End Date		31/12/2008
		
Total # of Trades		2232
Percent Profitable		83.83%
# of Winning Trades		1871
# of Losing Trades		361
		
Average Trade			$68.41
Average Winning Trade	$219.07
Average Losing Trade	$-712.47
Ratio avg. Win / avg Loss	0.31
		
Max. conseq. Winners		58
Max. conseq. Losers			7
Largest Winning Trade		$4660.00
Largest Losing Trade			$-2850.00
		
# of Trades per Day		0.76
Avg. Time in Market		349.1 min
Avg. Bars in Trade		329.3
Profit per Month		$1598.11
Max. Time to Recover	244.95 days
		
Average MAE		$278.68
Average MFE		$216.53
Average ETD		$148.12

Plus of course the ever-useful equity curve. Put the stalk of hay between your teeth, lean on the fence and say: "Don't like the look of that last little bit"

 

[Adamus]

Re: Gap Filler system - another equity curve for subjective

Plus of course the ever-useful equity curve. Put the stalk of hay between your teeth, lean on the fence and say: "Don't like the look of that last little bit"

Here's the same equity curve, to add a bit of objectivity to the subject. Perhaps displaying it this way will emphasise the retracements more and draw attention away from the desirable upwards slope, which is essentially irrelevant (and misleading since I haven't included any commission or slippage).

 

[Adamus]

Re: Gap Filler system - entering on a stop after a pullback away from the gap

If you the reader were reading this during the day today you'll see I deleted some posts. Just made a dumb mistake and have to re-do what I posted.

I'm backtesting with a limit entry instead of at the market at the open. I put the entry half way between the Open and the protective stop.

I'm expecting improvement to the avg win : avg loss ratio, and obviously with that the % win rate should drop from 80%.

If my theory's correct, making an entry closer to the protective stop loss should reduce the maximum loss.

If the entry is relatively near to the Open, then it should have the advantage of filtering out the trades that had a very poor risk - reward ratio. The disadvantage is in abandoning all those little profits.

- if the entry is relatively close to the protective stop, the trades are much more likely to be losers where the market just never turns back or at least, not before ploughing through the protective stop. The risk-reward ratio should be much improved, but the % winners will be ugly.

- the optimal place for the entry probably isn't right in the middle between the Open and the protective stop. Once I've found the optimal position for the basket of currencies as a whole, I'll have to compare the breakdown market by market.

Ideally, assuming that the optimal place for the entry is not right at the Open, then I'd like to see that all markets have improved uniformly. If the results per market have been messed up completely, i.e. some are just a nightmare and other are star performers, I'd have to look again at whether this is just curve-fitting.

Just got to figure out another little bug in my NinjaTrader code.

 

[Adamus]

Re: Gap Filler system - entering on a stop after a pullback away from the gap

I'm currently stuck with this Gap Filler system or better said, with the portfolio version of the Gap Filler system which gives me the full account level statistics for all the instruments traded in the backtests.

I have a bug or perhaps two bugs with the portfolio version which I can see by comparing the P&L between backtests using the same data, same period and same parameters.

I consider the normal NT7 single instrument version to be doing the correct trades, and the portfolio version isn't duplicating the results.

I get extra trades for the 'primary' instrument in the portfolio version - about 5 extra trades. The 'primary' instrument is just a technical artifact of NinjaTrader - it means the first instrument loaded, or the default instrument. You can't run a backtest without it.

Also in the portfolio version backtest, I am missing several trades for other instruments - it seems randomly - about 40. They are just missing.

But the trades that are present are exactly the same in the portfolio version backtest as they are in the single instrument backtests.

So the code which sets the stop and target sizes, which are based on the daily ATR, are fine.

It's the code which causes the system to enter or not that is up the creek. That's also based on the ATR, and on the point size of the instrument currently trading.

I suspect I tweaked the single instrument version and forget to add the code to the portfolio version. Not quite as dire a mistake as confusing stops and limits.

Right, I discovered one mistake. That's fixed the 40 trade discrepancies, but it still leaves the extra trades in the portfolio version due to the primary instrument.

This should be logical but I'll have to leave it now since I'm going out to celebrate Chinese New Year.

It's a shame since I wanted to get a final system sorted out today. Instead of searching for bugs, I didn't manage any backtesting at all. So another slow week gone. I did put some ideas together and manage a little testing but generally speaking development has been hideously slow again. I guess my expectations are as ever way too optimistic in terms of time management. I haven't even managed a planning session for this next week yet. OK so enough whinging - I have a portfolio strategy almost running now and hopefully this week I'll nail down a decent system in forward testing for Gap Filler.

I'll just run Gap Filler as it is tonight on the post-weekend open which should serve as good experience and might flush out some bugs so it's not a completely wasted opportunity.

I'll have to do the weekly planning tomorrow morning.

Gung Hei Fat Choi! - 4 days late, but nobody in Chinatown seems to let that worry them.

 

[Adamus]

Sunday afternoon planning session - rescheduled to Monday

Long term goals:

(1) Work out a specific figure that I want to make from trading at which point I will walk away from it and do something else
(2) Using my software development skills, build a trading platform that does all the good things I need that I know from the various different trading platforms I've used or evaluated, e.g. TradeStation, NinjaTrader, TradingBlox. Obviously way too big a job for one person, I'd like it to be an open-source collaborative project made up of trader-developers.

Medium term goals:

(1) Find or invent, write, test and implement trading systems firstly to cover monthly outgoings and secondly to increase the capital in my trading account. At this time this is a question of keeping my living costs as low as possible, and secondly putting as many systems as I can onto NinjaTrader. Since I have no idea where the limit for NinjaTrader lies, this is pretty much open ended. I can also set up a second computer anyway and buy another NinjaTrader license. The main bottle-neck on this is the rate at which I can produce systems.

(2) Software - current state: I have an application which can import trades/PnL originally for TradeStation but modified for any trade history, and run portfolio analysis against it. It also takes a TradeStation format order batch file and transmits them to IB TWS. I think my medium term goal should be to get this up and running without the database performance issue it currently has dealing with millions of trades.

The long term and medium term goals remain the same.

According to the 7 Habits system, I would have scheduled the short term goals to slots in my weekly calendar, but since I don't have a calendar I didn't do it. I do have a calendar, but it's on my ancient Palm Treo phone, and it's not suitable for the task really, it doesn't give me a good visual overview of the week. All the little reminders take up just as much space as the daily tasks would.

At least I can review what I did do from my short term goals and adapt them for this week.

Short term goals:

(1) Gap Filler system: run it tonight using either the 1 tick bars that allow me to buy or sell practically on the second/third tick after the IB open (so they say - probably the 10th tick knowing my internet connection) or run it using 1 mins bars that I use for doing the backtesting.

Ran it using 1-minute bars. Did that last night too, and still clearing the bugs out of the woodwork.

(2) Other low hanging fruit in the systems ideas list I have: Joel Rensink's One Night Stand and First Strike systems.

No progress

(3) Get some kind of definitive response from British Telecom on the network problems causing the disconnects in the IB TWS logs (that were responsible for missing $800 of profit last week)

Waiting for a call back - have been promised a call today from Friday.

(4) Software (must find a name for it): set up the project testing batch run and code a database performance test which fails if it falls below a desired level. That might take a bit of research.

No progress

(5) Sofware 2: get Apache and Tomcat running together on VPS

No progress

 

[Adamus]

Sunday afternoon planning session - rescheduled to Monday II

Short term goals this week:

(1) Gap Filler system: get it running for Sunday evening using 1 mins bars - debug the NT7 issues

(2) Gap Filler system: find a variation on the theme that produces the goods

(3) Other low hanging fruit in the systems ideas list I have: Joel Rensink's One Night Stand and First Strike systems

(4) Get some kind of definitive response from British Telecom on the network problems

(5) Software (must find a name for it): set up the project testing batch run and code a database performance test which fails if it falls below a desired level. That might take a bit of research.

(6) Sofware 2: get Apache and Tomcat running together on VPS

(7) Nail down the figures for the answer to long term goal (1)

So, in terms of time allocation, (4) & (7) are minor tasks.

I should spend a day at least on each of the remaining - (1), (2), (3), (5) & (6).

Conveniently that is 5 tasks for 5 days. Let's see if that works out next Sunday.

 

[Adamus]

Re: Networking!!!! Arrgh uugg heeaaawwwssssnnnnnnggngngng

But psychos aside, the dismal failure of Interactive Brokers historical data servers continues. Interactive Brokers' first advice to me was that the number of hops in my traceroute was too large, although they didn't tell me how to reduce it. It's currently 22:

traceroute to mktgw1.ibllc.com (208.245.107.9), 30 hops max, 40 byte packets
1 217.32.146.168 (217.32.146.168) 4.723 ms
2 217.32.146.238 (217.32.146.238) 6.444 ms
3 213.120.177.58 (213.120.177.58) 5.504 ms
4 213.120.176.58 (213.120.176.58) 5.691 ms
5 213.120.176.182 (213.120.176.182) 5.943 ms
6 acc2-10GigE-0-1-0-4.l-far.21cn-ipp.bt.net (109.159.249.198) 5.820 ms
7 core2-te0-15-0-2.ealing.ukcore.bt.net (109.159.249.161) 7.055 ms
8 transit2-xe1-1-0.ealing.ukcore.bt.net (62.6.200.142) 6.090 ms
9 t2c2-ge13-0-0.uk-eal.eu.bt.net (166.49.168.53) 6.316 ms
10 195.50.91.153 (195.50.91.153) 6.639 ms
11 ae-32-52.ebr2.London2.Level3.net (4.68.117.62) 11.097 ms
12 ae-3-3.ebr1.London1.Level3.net (4.69.141.189) 6.955 ms
13 ae-100-100.ebr2.London1.Level3.net (4.69.141.166) 7.403 ms
14 ae-41-41.ebr1.NewYork1.Level3.net (4.69.137.66) 75.892 ms
15 ae-4-4.ebr1.NewYork2.Level3.net (4.69.141.18) 76.024 ms
16 ae-1-51.edge2.NewYork2.Level3.net (4.69.138.195) 75.763 ms
17 4.68.110.70 (4.68.110.70) 89.762 ms
18 0.ae3.XL3.NYC4.ALTER.NET (152.63.16.182) 74.413 ms
19 0.ge-7-0-0.XL3.BOS4.ALTER.NET (152.63.0.165) 82.597 ms
20 POS6-0-0.GW12.BOS4.ALTER.NET (152.63.22.177) 82.994 ms
21 interactivebrokers-gw.customer.alter.net (208.192.181.62) 90.656 ms
22 mktgw1.ibllc.com (208.245.107.9) 90.989 ms

It used to be 26 before I added OpenDNS and Google DNS servers to the DNS list.

A friend of mine who has Virgin cable ran the same traceroute and only has 9 hops. How does he manage that?!!?? Everyone says BT is **** but I never actually believed them until now.

BT is on notice that I intend to cancel my contract with them due to this continual networking breakdown.

Now more than just connecting to the historical servers, I examined my logs and discovered that there is some sort of witching hour between 4am and 5am where the live server connection breaks repeatedly, sometime 10 times I've counted. Fortunately both IB and NT7 recover from the disconnects without me noticing.

To tell the truth I did notice that there were often disconnects overnight because I scan the logs in the morning and there are repeated occurences of the disconnect warnings. However since last week when the disconnects kept on occuring right up until 6:30am, I hadn't worried about it. Now I do though. A disconnect happened right at the point when NT7 was submitting the target and stop orders for the FMT system on the GBP. NT7 lost track of the orders, paniced and killed FMT and flattened the position. This happened 2 days in a row which both would have been winners - i.e. it cost me $800 in foregone earnings.

So I have 2 complaints for BT. The original disconnects occur from 14:00 to 23:00 on the mktgw1.ibllc.com server in NY, and the early morning disconnects occur on the mktgw1.ibllc.ch in Switzerland.

I'm quite eager to find out what they say although my gut feel is that they'll change nothing and give some half-baked excuse and I'll terminate my account with them.

I think Virgin Broadband is the best on offer from a quality point of view and I have access to a friend's LAN where I can set it up for testing overnight.

There is also the strong chance that I'll rent a server in NY anyway to give me the freedom to be away from my desk but still run the systems.

I'll also look at doing that with VPN for security's sake.