{% extends "global/Page.html" %}
{% load otree static %}

{% block title %}
    Instructions
{% endblock %}

{% block content %}


    <br/>

<p>
    <strong>The task of this experiment is to make forecasts for the future demand of a product, with the objective to be as
    accurate as possible. The accuracy of your forecasts affects your final payout.</strong>
</p>


        <p>
            You are part of a company and your role is to <strong>forecast the future demand of a product</strong>. You don't have to
            care about how much of the product you have in stock or if you would have the capacities in production
            to meet the demand. You merely make a prediction of what the actual future demand will look like. Others people
            in your company can make use of that information and the more accurate your prediction, the more profitable the company.<br/>


        </p>

        <p>
            Every day, you make a forecast for the upcoming day lying ahead. There are
            <strong>two decision stages</strong> you have to go through for making a single forecast. Only after going
            through both decision stages the forecast for the next day is set and you move on to the forecast for the
            day after.<br/>
            In a <strong>first decision stage</strong> you make a forecast. The <strong>entire historical data</strong> of the product's
            demand your company has collected is available for you to inspect. The time series of historical data is <strong>created from real data</strong>.<br/>
            In a <strong>second decision stage</strong> you get the opportunity to revise your forecast and
            make a final forecast. Next to historical data, there is a <strong>forecast of a machine learning (ML)
            algorithm</strong> available. The ML algorithm is based on a model, created by experts.
            The entire historical data you can see and inspect is <strong>exactly the same</strong> as the data available to the ML algorithm.
            The ML algorithm was trained on four years of data.<br/>
            In both decision stages you can enter a numerical value between 0 and 1000 as your forecast. In total you will make forecasts
            for 30 days. The days you already made a forecast for become part of the historical
            data.
            <br/>
            After you finalized the forecast for one day you will see a <strong>round summary</strong> displaying your initial (first decision stage)
            and final (second decision stage) forecast as well as the ML algorithm forecast and the actual demand value for the day you made a forecast for.
        </p>


        <h4>Payout</h4>
        <p>
            At the end of the experiment 5 forecasting values (i.e. 5 days) will be drawn randomly from all your
            final forecasting values. Their average accuracy determines the final accuracy. After all experiment
            sessions have been conducted, the top 5% of the participants based on final accuracy will receive
            a price.<br/>
        </p>


    After pressing the "Next" button, <strong>you have to answer 6 understanding questions<strong> before moving on.
    {% next_button %}

{% endblock %}