Building a Network

Configuring the tabs sets how the network learns; building the network defines what it learns from and predicts. You do that by populating the tree with inputs, outputs, optional code, and — for a Standard network — a pool of symbols to train across.

A populated network

Here is a small but complete example network, "DemoNet": two inputs, one output, a helper code snippet and a three-symbol training pool. With a name, two inputs and one output, the network is valid and Generate Formula turns on.

1. Neural Network Name — set to DemoNet; names the AFL section and the saved .net file.
2. Inputs — the RSI and 30-day EMA inputs added via templates.
3. Outputs — the prediction target the network learns.
4. Global Code — the reusable AFL snippet (Helper).
5. Ticker Training Pool — AAPL, MSFT, SPY to train across.
6. Generate Formula — enabled once the network is valid (name + 2 inputs + 1 output).

What to add

Each part of the tree has its own page in this section:

• Inputs — the indicators or price series fed into the network, with lag and percent-change options.
• Outputs — the value to predict and how far ahead, with plotting options.
• Code snippets — custom AFL inserted into the generated formulas.
• Ticker training pool — the symbols a Standard network trains across.

Once the tree is populated, head to Generating the Formula.