Cycle 4, stage 4, day 14:
I am still working on bugs. I was working on the last one last night. On my way on fixing it too as well.
While I was sleeping, a crash did occur. I have been lucky, when the system crash in the middle of a trade, it is sometime messy to manually unwind what the system was in the middle of several trades. The system had the time to clean everything before crashing leaving the account balances in a clean state.
It sucks fixing bugs but I consider them as good luck. I have been lucky to have the perfect conditions to make the bugs show themselves. In all the recent occurrences, they did happen because new code has been triggered following some rare or exceptional situations. Once I clean those bugs, the system should become rock solid.
I needed money and I was wondering where I would find it. I did receive a notification that finally, stacked ETH2 can be unstacked this morning...
Cycle 4, stage 4, day 15:
I have almost put in place another feature that I have been working for a long time without never completing it as there was always something more important to do than completing it.
I was on the verge on trying it. Everything seem to be working fine... Except a minor shortcoming in the GUI. GUI dev is not my strongest skill and this made me stuck... After having sleep in the issue, I am pretty confident that I have a good plan to finish the thing...
However, the first thing that I did this morning is how the trading system went during the night. It did ok but it did stump into another new glitch. It never stops to amaze me in how different ways the system can fail... I guess that this is predictable given the size of the project. I have just counted the lines of code that I have in 2 folders... The 2 folders contains over 300,000 lines of code... With the other folders. The project must contain over half a million lines of code... This is for sure my biggest project of my life...
Anyway, I was puzzled by what I was observing and I have finally been able to find the problem...
(04-15-2023, 06:47 AM)lano1106 Wrote: [ -> ]Cycle 4, stage 4, day 15:
I was on the verge on trying it. Everything seem to be working fine... Except a minor shortcoming in the GUI. GUI dev is not my strongest skill and this made me stuck... After having sleep in the issue, I am pretty confident that I have a good plan to finish the thing...
Though I have no doubts in your technical abilities, I can not overstate the benefits of bouncing your ideas off of ChatGPT (chat.openai.com). Still free. I work as a web developer and it's blown up my ability to get through the syntax and focus more on the concepts and features of the code I'm working on.
For context, ChatGPT will interpret and breakdown your code errors, convert your idea/prompt into actual code (it assumes Python, but feed it your language in your prompt), and help you expand your knowledge of the tools or languages you are already using.
The quality of the response depends on the quality of the question. Meaning that you need to add enough detail and context for it to take it from there. But I'm telling you, it's like asking a senior developer a question, as it gives you a conversational walkthrough of new concepts and of code. Be aware: I've been told that some of the code suggestions might complicate debugging later on, but this can be mitigated by enough counter-measures (unit tests, architectural knowledge, etc.)
oh thx... I am not too sure if I want to try AI because I assume that it can be extremely addictive... I did not realize that it could assist in programming as well... well thx for the suggestion... I might give it a shot and see if it can help me making things happen faster...
For now, I have been able to address 100% my UI issue by myself... I am super happy. My idea did work perfectly on first try and it is even better than what the code was doing before...
The funny thing is that by being unblocked with new possibilities did make the code see a new situation that did never happen before and this did result into a core dump of my GUI client. It took about half an hour to figure out the bug and I have finally nailed it...
This seems to be the theme of the week-end... New bugs are showing up their ugly faces... They are all gone... My codebase is maturing seriously.... the day is pretty much over... I am going to do something fun with the kids for the rest of the evening...
So implementing my new alpha model is going to wait until tomorrow before I start it... I am crossing my fingers to not have any other bugs to take care of.... Markets are extremely quiet today... Traders are probably tired of all the activity that did happen in the last few days.... I have high hopes that my latest idea is going to generate a very significant income increased... My wildest dreams are telling me that it could generate few Ks daily...
Cycle 4, stage 4, day 17:
if there is something that can happen, it will happen.
Few days ago, I have addressed a very exceptional asset balance management issue. By its exceptional nature, having the opportunity to try out the fix was not immediate. It did turn out that there was a regression bug in the fix and it manifested itself yesterday morning.
It took me the whole day yesterday to find it. I was doing insomnia due to my obsession to understand the issue... That has been a hard nut to crack. I had few hypothesis to explain my observation but the logs were not supporting them.
It is funny how when you focus on a certain part of a system, you keep discovering imperfections... I have cleaned and polished few minor things but not the one that I hunting down... I was about to make insomnia despite being tired last night. I could not sleep, so I went back to my computer to keep looking for the cause of my issue and I have finally found it... It was one of the hypothesis that I have imagined... I had modified the code to address the theoretical issue but I was not sure that it was the one... I have finally stumbled into a log trace confirming that my suspicion was correct.
I was too tired to fix it right away but it has been an immense relief to know that the mystery was resolved. I have completed the fix few minutes ago... I am not 100% satisfied of the result as it is kinda complex and not as elegant as I like but this is the best that I can think of to manage a complex race condition... I have left plenty of documentation in case that this issue would reappear down the road...
So, I am ready to close the current issues and glitches chapter and hurry up the implementation of my latest idea that might be the game changer feature that I am working to find...
Update:
A new problem did appear... Hopefully, it is going to be easy to fix. I think it will...
Cycle 4, stage 4, day 18:
the other problem took pretty much the rest of the afternoon to fix and the remaining of my available mental energy to create code. I might also have seen another issue...
I was feeling overwhelmed with those issues. Well the first one is a valid issue. It was in the system since the week-end when I did enable my new multi-account management feature. I am just stunned about the timing that it shows up. Possibly in the worst possible moment. It could have appear before but it decided to show up precisely yesterday... I guess that there is no good or bad timing... Life is surprising...
The second issue... I am now confident that the problem is coming from the exchange ended up being mixed and pushing me back old updates when I reconnect to it which is messing with my system... I am not 100% sure how to handle this disruptive exceptional situation but I have a good idea... I have improved my logs to catch the next occurence and confirm that my idea might be the good one...
well, for now, I have been able to put back the cover on top of the bugs and issue source hole... I am hoping for a productive day today...
As a side note, last night was my night off... It has been filled with a lot of dreams... It is as if... when playing the program, the subconscious mind is listening and during the nights off, it express itself...
Cycle 4, stage 4, day 20:
I have finally completed my first prototype of my new alpha model that I hope is a game changer in my journey...
Actually, it is an old new idea... A recurrent idea that I first had in November 2020 but back then I had not all the knowledge and infrastructure in place to actually implement it.
So far, it has not been triggered... So many unknowns... I don't know if the initial draft is correct and contains no bug... I don't know yet how often it can be potentially be triggered.
Is it once per hour? (I now know that this is no)
Is it once per day? per week? per month?
I'll need some time to discover its characteristics. Maybe I need to finetune it....
So there is a lot of things that I am not sure yet... but I am pretty confident that when it is going to be triggered, it is going to be strong signal that should be very profitable... time will tell...
I have at least 3 additional improvement layers that I could add over the initial prototype but before going ahead, I'll wait to let the prototype show a little bit its potential...
In the meantime, this is a new automated way to make money that is in place... I am now available to work on the next method to achieve UMS...
Cycle 4, stage 4, day 21:
Last night, my new model got triggered. It turns out that the signal was a false positive on its first report and by being activated, this did allow me to find out that the code was not 100% correct.
I did fix the discovered glitch and I may need to do some more finetuning but preliminary result is not bad...
Cycle 4, stage 4, day 23:
My new model has started to work. I had to correct a bunch of small problems such as closing too soon a winning position...
but I am impressed by the result... Sometime, it detects an opportunity and open a position and because it is under close scrutiny, I go look at the price chart to see if the move makes sense... This morning when it did happen, I concluded that the model made a mistake and detected a false positive but just for the sake of testing the rest of its position management, I am letting the position open only to find out that opening the position was an amazingly good idea that I could simply just not detect myself by looking at the charts...
I am keeping optimizing the model code and I have only let it trade with a $25 budget per position but I am very close to make it trade with a much bigger budget... and for my sanity, I just let it run up to 2 positions simultaneously... I have increased that limit to 4!
It is a bit less exciting than the instantaneous arbitrage result... but the suspense is thrilling as the price is moving very slowly toward profit... I am so glad to have coded the trading strategy... I would have a hard time staying calm and taking rational decision... Right now, I don't have to touch anything...
I feel it is coming...
If I look back one month ago. If there is 1 crucial event that have tipped the balance toward my current situation. I would say that it is the 2 days electric blackout that has broke my thought process and my current pattern and it did force me to find other goals to pursue. This has been a crucial moment. If this did not happen, my effort direction would not have steered in the direction that I went in the last 2-3 weeks...
I am forced to say that there must be some luck in the past events... but you have to receptive to opportunities to grasp them... I think that my 4th UMS cycle is going to be the good one...
Cycle 4, stage 4, day 27:
My celebration of almost getting to success in the last entry was premature...
The secret sauce is still not very good... Despite improving a lot of things in the system, it seems like the initial success was in part due to luck and it was not something that could be systematically repeated...
Not everything is bad... I probably have few ideas on how to modify the model to make it succeed more predictably...
I will play with those ideas in the new few days...
Cycle 4, stage 4, day 28:
I did track down another nasty bug... The nastiest bugs are those that stay dormant in the most regular usage and show up when something out of the ordinary happens.
It took the whole day to figure this one out... I have been thinking about letting it go and move on... that is not me doing something like that.
In the end, once I have figure out what was the problem, I am super glad to have gone to the trouble of finding out the issue... It was a serious problem and now it is fixed!
Cycle 4, stage 4, day 29:
Beside the nasty bug that I have mentioned in my last entry... Searching its cause made me find another bug. Today, I have returned to improve my strategy decorator. I had a new improvement in mind but how to do it was boggling my mind... I did put in place a simple decorator system to decorate the order strategies to add some temporary functionality to them...
The new functionality was to be applied to a sequence of strategies... My initial approach I had was to create a new strategy class which would have been the concatenation of all the strategies in the sequence with a finite state machine that dispatch its calls to the right substrategy. It was a complex mess...
I got a ah ah moment where I have figured out that if only my decoration system could:
Stack several decorators
Transfer the decorators when the strategy is replaced to the new one,
this would allow me to only write a new decorator instead of creating a complex mess... I did this today and I am quite satisfied of the result!
However, upon restarting my server, the little devil started to crash... I found another problem. My fault... I must have introduced this bug last week during some refactoring that I did for my new alpha model... The bug needed some very special conditions to show up which did happen this afternoon....
The core dump was making little sense because of optimization.... I have finally figured out what was the problem and I did fix it. As I am writing this entry the system is priming up and should start working in few minutes...
To summarize. 3 bugs fixed and a new feature has been added... Not bad!
Cycle 4, stage 4, day 30:
my alpha model has flaws. it did turn out to be not as good as I did imagined...
I guess that the only way that I could figure that out was to give it a shot.... or else I would have stayed with the thought that I left on a drawing board a great idea
that being said, I have learned stuff from the experiment and as a result, a new spin on the current design did pop out in my mind...
Cycle 4, stage 4 (month 2), day 2:
Yesterday has been a hard day... I have been struggling to complete a modification of my latest model for it have a better performance...
I have finally completed the task this morning... I have launched it... and I went to the gym... I was a bit worried to leave a major change run unattended without validating that it was working fine. What had to happen did happen. This is that, the code has hit a deadlock...
It took me maybe an hour to figure out the problem and write a fix. The fix is nice. The bug did make see small other glitches that went unnoticed until I have to investigate this part of the code....
I have launched the fixed version... I went to eat and when I came back, I have found my server core dumped again...
this time, it is because the exchange has launched the trading of new stablecoin. The EURT... because my client does not know this currency, it is messing with my parser when it process pairs with EURT in it.
It mistakenly return the lexer token for the fiat EUR symbol instead...
It is not a too big deal to fix this second problem... but I am feeling resistance from the universe when I am eager to try out my latest model modification which I think will perform better...