WEBVTT 00:00.200 --> 00:03.520 And now we move on to our trader. 00:03.960 --> 00:08.360 So our trader we're going to give our trader a strategy. 00:08.360 --> 00:15.400 So that I'm going to name our traders by the standard cryptocurrency names that they give things beginning 00:15.400 --> 00:16.200 with A, B and C. 00:16.240 --> 00:20.200 So Alice is our first trader and uh, yeah. 00:20.200 --> 00:24.440 So, so, um, we begin by giving Alice a trading strategy. 00:24.480 --> 00:30.200 You are a day trader that aggressively buys and sells shares based on news and market conditions. 00:30.720 --> 00:34.360 And we're going to use account get Alice. 00:34.400 --> 00:41.880 And we're going to call a method reset on account that resets Alice's strategy to this strategy here. 00:41.880 --> 00:48.000 And it also takes her money to exactly 10,000 to a starting position. 00:48.000 --> 00:49.920 So this gets us to our starting point. 00:49.920 --> 00:52.120 And obviously we're calling the account code directly. 00:52.120 --> 00:55.120 There's no point in using an MCP tool for this. 00:55.120 --> 00:59.520 This is just us getting things ready, getting our environment prepared okay. 00:59.560 --> 01:02.880 And then once we've done that I will use MCP to show you. 01:02.920 --> 01:06.000 We will use it to call reading the resources. 01:06.000 --> 01:07.720 And there are actually two resources. 01:07.720 --> 01:14.440 There's one for looking at Alice's account, and there's one for looking at her investment strategy 01:14.440 --> 01:15.960 that we've hopefully just set. 01:16.000 --> 01:19.720 When we call these, it will be it'll be in the account codes. 01:19.720 --> 01:20.760 It will be in here. 01:21.080 --> 01:30.320 Uh, when we call reset, it's going to uh it's going to both um, uh, create Alice's, uh, account 01:30.320 --> 01:35.160 and set it up, but it's also going to then save it in into to the database. 01:35.160 --> 01:39.560 So if we look at the reset method right here, it sets the balance to the initial balance. 01:39.560 --> 01:46.000 That's the strategy to the strategy wipes out the holdings and the time series and then saves and save 01:46.040 --> 01:52.800 is going to save uh everything down to our SQL, our SQLite database. 01:53.320 --> 01:56.880 So let's run this and see if it all works. 01:58.240 --> 02:00.120 Well it did and it was very fast. 02:00.120 --> 02:07.400 So we've got a trader called Alice who's got 10,000 in the bank, no holdings, no transactions and 02:07.400 --> 02:09.160 a total portfolio value of 10,000. 02:09.200 --> 02:11.320 And that is Alice's investment strategy. 02:11.360 --> 02:14.040 A day trader aggressively buys and sells. 02:14.040 --> 02:16.080 So, so far, so good. 02:16.280 --> 02:19.520 And now time to create our trader agent. 02:19.680 --> 02:20.800 Okay, so this is it. 02:20.800 --> 02:23.280 This is the agent for for Alice. 02:23.480 --> 02:28.720 So the account details is that we are going to read that from the resource. 02:28.720 --> 02:29.640 Same for the strategy. 02:29.680 --> 02:31.320 We put together some instructions. 02:31.320 --> 02:32.440 The system prompt. 02:32.480 --> 02:35.160 You're a trader that manages a portfolio of shares. 02:35.200 --> 02:36.560 Your name is Alice. 02:36.600 --> 02:38.400 Your account is under the name Alice. 02:38.800 --> 02:44.520 You have access to tools to search the internet, to check stock prices, to buy and sell shares. 02:44.680 --> 02:50.720 Your investment strategy is this and your account holdings and balances this your tools. 02:50.720 --> 02:59.000 To do this, we should also say we should add in here you have tools to, uh, save. 03:00.050 --> 03:01.530 a memory. 03:03.610 --> 03:06.370 Of companies. 03:07.610 --> 03:12.290 Research and thinking. 03:12.890 --> 03:14.010 So far. 03:14.170 --> 03:20.610 So we'll give Alice some knowledge of the fact that Alice has the the persistent tools as well. 03:20.850 --> 03:23.010 Use your tools to make decisions about the market. 03:23.010 --> 03:23.650 Investigate. 03:23.650 --> 03:27.930 Make your decision, make the trades and respond with a summary of your actions. 03:28.210 --> 03:29.010 That's great. 03:29.050 --> 03:31.890 Let's just see that printed out now with with the stuff in it. 03:31.930 --> 03:36.530 And what you'll see is that the current holdings and the balance have been written in there. 03:36.770 --> 03:41.370 And you'll see that the strategy is in here as well. 03:41.610 --> 03:42.610 There it is. 03:42.850 --> 03:45.490 So it's all clearly laid out. 03:45.770 --> 03:48.970 And now it's time for us to run our trader. 03:49.810 --> 03:54.850 So this here is going to connect with all of our servers. 03:54.850 --> 03:56.490 And actually we can just write that simply. 03:56.530 --> 04:02.650 We can make that for server in MCP servers await server connect. 04:02.650 --> 04:07.250 So that will just connect all of our servers so that they should all be ready to go. 04:08.330 --> 04:15.890 And we will now call the to get the researcher tool passing in the researcher servers. 04:16.530 --> 04:23.410 And then we are going to create a trader called called called Alice. 04:23.970 --> 04:28.050 Going to pass in that system prompt those instructions for the tools. 04:28.050 --> 04:35.770 We're going to pass in the trader tools, which is either an empty list or it's a list of the free tools 04:35.770 --> 04:36.730 for market. 04:36.770 --> 04:39.130 If you if you're going with the free approach. 04:39.530 --> 04:42.490 And then we're going to add in our researcher tool right here. 04:42.770 --> 04:45.170 We're going to pass in our MCP servers. 04:45.410 --> 04:49.610 And we're going to then uh runner dot run. 04:49.610 --> 04:55.890 That is of course the way that we kick off the entire run, including delegating to the tool to call 04:55.890 --> 05:00.890 the researcher and putting all of the results back in the variable result. 05:00.930 --> 05:02.610 Okay, let's see what happens. 05:03.730 --> 05:04.650 Off it goes. 05:06.490 --> 05:07.970 It's going to think about that. 05:08.570 --> 05:11.090 While it's doing that I might just set. 05:11.130 --> 05:16.210 Maybe we should uh right after this we would want to print the, um. 05:16.250 --> 05:16.610 Oh, sorry. 05:16.610 --> 05:17.290 I've got that already. 05:17.290 --> 05:18.210 We're displaying it right here. 05:18.210 --> 05:23.890 Well, then maybe right afterwards we'll want to go and have a look at our, uh, at our, um, trace. 05:29.010 --> 05:34.650 I'll do that while it's thinking, which is a platform dot OpenAI. 05:39.530 --> 05:41.050 Okay, well, here we go. 05:41.090 --> 05:43.850 It completed, uh, market overview. 05:43.850 --> 05:49.010 The market is experiencing significant downturns, with major indices declining due to tariffs, uh, 05:49.290 --> 05:50.730 which is happening right now. 05:50.930 --> 05:52.850 Here are some stock price movements. 05:52.930 --> 05:55.050 And so it's made some trading activities. 05:55.050 --> 06:02.130 It claims that it has uh Purchased ten shares of Netflix tried to sell ten shares of Apple. 06:02.210 --> 06:03.410 Well, that seems a bit silly. 06:03.450 --> 06:08.570 It says it didn't hold any shares, so no doubt it's no surprise that it complained. 06:08.570 --> 06:10.050 Maybe it was trying to go short. 06:10.330 --> 06:13.930 Uh, maybe we should say that you're not allowed to sell shares you don't have. 06:14.210 --> 06:15.690 Um, and, um. 06:15.730 --> 06:16.290 There we go. 06:16.330 --> 06:18.450 Then this is the summary at the end. 06:18.810 --> 06:21.770 So let's, uh, first of all, see if we agree with that. 06:21.770 --> 06:24.490 Let's read the accounts resource and see what's happened. 06:24.650 --> 06:29.730 So it does indeed at this time hold ten Netflix shares. 06:29.730 --> 06:31.410 So the trading decision worked. 06:31.410 --> 06:34.970 It did buy ten shares of Netflix just as it thought it did. 06:35.330 --> 06:39.250 And those are the transactions that it's that it's made right there. 06:39.690 --> 06:46.370 And uh, it's got the rationale in here that it's written against that and the portfolio value and the 06:46.370 --> 06:53.090 trading loss, a tiny trading loss because of the trading fees, the the bid offer spread. 06:53.370 --> 06:54.810 Um, and so there we go. 06:54.850 --> 07:02.420 We've just seen a successful trading run decisions made and we've seen many MCP tools being used. 07:02.420 --> 07:05.180 It's used the tools to search the web. 07:05.180 --> 07:09.220 It's used the financial tools to look up share prices and financial news. 07:09.220 --> 07:13.100 And then it's used our tools to actually make the trade. 07:13.100 --> 07:16.900 So let's go and look at the trace now in OpenAI. 07:17.620 --> 07:19.180 So here it is Alice. 07:19.820 --> 07:20.700 Because we did that. 07:20.740 --> 07:22.420 The 13 different tools. 07:22.820 --> 07:23.980 Uh wow. 07:24.020 --> 07:25.220 Let's take a look at what happened. 07:25.220 --> 07:26.900 So there's a lot of lists that went on. 07:26.900 --> 07:32.060 So first of all the researcher was called the researcher, which was in there as a tool. 07:32.060 --> 07:33.740 It was collaborating as a tool. 07:33.940 --> 07:40.060 And the researcher also, of course, listed its tools, and it did a total of three brave searches. 07:40.340 --> 07:44.700 It looked for the current market top performing stocks. 07:44.700 --> 07:46.180 It did as that's great. 07:46.700 --> 07:50.180 Current Market Trends Stocks April 2025. 07:50.340 --> 07:55.980 So it did a bunch of of, uh, of web search queries to gather that information. 07:56.300 --> 08:02.020 Um, and you can see that it's it's definitely it's taking the brave search in favor of just trying 08:02.020 --> 08:04.180 to retrieve web pages itself. 08:04.540 --> 08:09.540 Um, and you can see there's we have to press load more to get more so we can see that. 08:09.540 --> 08:12.740 Then the trading agent went and got lots of stock prices. 08:12.780 --> 08:18.540 Amazon, Apple, Microsoft, Tesla, Google and Netflix. 08:18.540 --> 08:19.940 It got all of those. 08:20.180 --> 08:23.300 And then it decided to buy some shares. 08:23.300 --> 08:25.460 It bought Netflix shares. 08:25.500 --> 08:28.340 And then just as it said it tried to sell Apple. 08:28.340 --> 08:33.620 And it got the error back and it tried to sell Tesla and got the error back, and then it crafted its 08:33.620 --> 08:35.100 response back to us. 08:35.300 --> 08:39.300 So the trace is where we can go and see everything that happened. 08:39.340 --> 08:46.420 See all of the activities that our trader, uh, carried out and see it's sort of inner thoughts as 08:46.420 --> 08:47.180 it did it. 08:47.340 --> 08:53.460 And so this is a super useful to see what's going on behind the scenes and to confirm that our trader 08:53.460 --> 08:57.620 and researcher are collaborating and working exactly as we planned.